THE ORIGINAL AND MOST POPULAR AROUND THE WORLD

SUCCESS
SERIES

THE A TO Z OF MACHINE LEARNING

YOUR ULTIMATE MCQ EXAM PREP RESOURCE

DESIGNED FOR KNOWLEDGE TEST

MCQ EDITION

NARAYAN CHANGDER
 Coverage of all important topics.

 The excellent aid for better rank.

 Perfect for all competitive exam.

 Use this book with your main text.

 Toppers 3FIRST3 choice I G

B RD
 Over 1060+ questions. M A
E A H
R K
D R
O
W
USEFUL FOR

□STUDENTS ✔
✔ □TEACHERS ✔
□RESEARCHERS ✔
□MIDTERM EXAM
□ENTRANCE EXAM ✔
✔ □COMPETITIVE EXAM ✔
□QUIZZES ✔
□OTHER
 Disclaimer

The aim of this publication is to sup-

ply information taken from sources be-
lieved to be valid, reliable and authen-

NARAYAN CHANGDER
ticate. The author bear no responsibil-
ity for any damage arising from inad-
verent omissions, negligence or inac-
curacies (typographical or factual) that
♣
may have found their way into this PDF
booklet.
Due care has been taken to ensure that
the information provided in this book
is correct. Author is not responsible
for any errors, omissions or damage
arising out of use of this information.

nt
Importa inter-
s , s e a rch the de
er to inclu -
i t h t h e answ w a n t
w u au
atisfied . If yo ontact
If not s rrect answers klet, please c t s:
p
net for
co
i n t h is boo F a c e b ook ht
estions on
tact him arayanchangd
er/
new qu a n c o n n
ou c om/
thor. Y acebook.c
. f
//www
 Contents

1 Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Machine Learning and applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Types of Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.4 Supervised Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2 Application of supervised learning . . . . . . . . . . . . . . . . . . . . . . . . 23

2.1 Unsupervised learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 Supervised and unsupervised learning . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3 Classification in Machine learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 Support vector machine (SVM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5 Reinforcement Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.6 Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.7 Gradient Descent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.8 NEURAL NETWORK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.9 Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.10 Perceptron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.11 Machine Learning Pipeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.12 Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
2.13 Machine learning(HARD QUESTIONS . . . . . . . . . . . . . . . . . . . . . . . . . 113

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
NARAYAN CHANGDER
 1. Machine Learning

1.1 Introduction
1. Which of these statements is (A) Linear regression
true about classical ML vs. (B) Logistic regression
deep learning?
(C) Gradient Descent
(A) All deep learning algorithms
are machine learning algorithms (D) Greedy algorithms

4. Generally, which of the fol-

(B) All machine learning algo- lowing method(s) is used
rithms are deep learning algo- for predicting continuous de-
rithms pendent variable?
(C) Classical ML is a subcate- (A) Linear Regression
gory of deep learning algorithms,
(B) Logistic Regression
based on neural networks
(C) Both
(D) none of above
(D) None of these
2. What will be the value
of F-1 score, if precei- 5. A computer program is said
sion=recall=0.5? to learn from E with re-
spect to some class of tasks
(A) 1
T and performance measure
(B) 0.5 P, if its performance at tasks
(C) 2 in T, as measured by P, im-
proves with E.
(D) None of these
(A) Training
3. In which technique you (B) Experience
can depict a relation be-
tween two variable using a (C) Database
straight line (D) Algorithm

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
4 Chapter 1. Machine Learning

6. What is “ML”? 11. Suppose, you designed a

model to predict grades
(A) Maternity Leave
of a student in a subject.
(B) Mixed Language Which metric will you use
(C) Missing Link to measure accuracy of the

NARAYAN CHANGDER
model?
(D) Machine Language
(A) F-1 score
7. has been used to train ve- (B) Preceision
hicles to steer correctly and (C) Recall
autonomously on road.
(D) Mean square error
(A) Machine Learning
12. A broad term that refers
(B) Data mining
to computers thinking more
(C) Neural Networks like humans.
(D) Robotics (A) Artificial Intelligence
(B) Machine Learning
8. Open-source tool that can
combine code, markdown, (C) Deep Learning
and visualizations together (D) none of above
in a single document.
13. This type of learning to be
(A) Visual Studio used when there is no idea
(B) Visual Studio Code about the class or label of a
particular data
(C) Jupyter Notebook
(A) Supervised learning algorithm
(D) none of above

9. Machine learning is an ap- (B) Unsupervised learning algo-

plication of: rithm

(A) Deep learning (C) Semi-supervised learning al-

gorithm
(B) Artificial intelligence
(D) Reinforcement learning algo-
(C) Expert systems rithm
(D) Robotics 14. The Q-learning algorithm is
a
10. Field of study that gives
computers the ability to (A) Supervised learning algorithm
learn without being explic-
itly programmed. (B) Unsupervised learning algo-
(A) Deep Learning rithm
(C) Semi-supervised learning al-
(B) Machine Learning
gorithm
(C) Data Science
(D) Reinforcement learning algo-
(D) none of above rithm

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.1 Introduction 5

15. Let’s say we have an im- (A) Artificial Intelligence

age of size 45x55. It is
(B) Machine Learning
a grayscale image. How
many pixels do we need to (C) Deep Learning
represent this image in a nu- (D) Traditional Learning
merical format?
(A) 2240 20. How many types are avail-
(B) 2420 able in machine learning?
(C) 2152 (A) 1
(D) 2340 (B) 2
16. Which one of them finds hid- (C) 3
den structures in data (D) 4
(A) linear regression
(B) Supervised learning 21. Real-Time decisions, Game
AI, Learning Tasks, Skill
(C) UnSupervised learning Aquisition, and Robot Nav-
(D) none of above igation are applications in

17. Adding a non-important

feature to a linear re- (A) Unsupervised Learning:Clustering
gression model may
result in.1.Increase in (B) Supervised Learning:Classification
MSE2.Decrease in R-square
(A) 1
(C) Reinforcement Learning
(B) 2
(D) Unsupervised Learning:Regression
(C) 1 and 2
(D) none of these 22. Machine Learning can be
classified into the following
18. If a linear model is fitted accept
on a non-linear data, then
what will be the value of (A) Reinforcement Learning
bias? (B) Unsupervised Learning
(A) High (C) Deep Learning
(B) Low
(D) Supervised Learning
(C) Average
(D) None of these 23. What does Machine Learn-
ing means?
19. is a category of an
algorithm that allows soft- (A) Teaching the machine
ware applications to be- (B) Learn & adapt
come more accurate in pre-
(C) Train the data
dicting outcomes without be-
ing explicitly programmed. (D) Make machine like human

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
6 Chapter 1. Machine Learning

24. What type of machine made to simplify the train-

learning algorithm makes ing process)
predictions when you have (A) Variance
a set of input data and
you know the possible re- (B) Overfitting

NARAYAN CHANGDER
sponses? (C) Underfitting
(A) Unsupervised (D) Bias
(B) Reinforcement 29. In a simple linear regres-
(C) Supervised sion model (One indepen-
dent variable), If we change
(D) Deep Learning
the input variable by 1 unit.
25. How many coefficients do How much output variable
you need to estimate in will change?
a simple linear regression (A) By 1
model (One independent (B) No change
variable)?
(C) By intercept
(A) 1
(D) By its Slope
(B) 2
30. A model of language con-
(C) 3
sists of the categories which
(D) Cannot Say does not include

26. Different learning methods (A) Language units

does not include (B) Role structure of units
(A) Memorization (C) System constraints
(B) Analogy (D) Structural units
(C) Deduction 31. What year did Deep Learn-
(D) Introduction ing breakthroughs drove
A.I. (Artificial Intelligence)
27. The difference between the boom?
average prediction of the (A) 2010
model and the correct value
which we are trying to pre- (B) 2009
dict is called as (C) 2001
(A) Variance (D) 2008
(B) Bias 32. In terms of the bias-
(C) Loss variance trade-off, which
of the following is/are sub-
(D) None of these
stantially more harmful to
28. Error that results from in- thetest error than the train-
accurate assumptions in ing error?
model training (that are (A) Bias

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.1 Introduction 7

(B) Variance 37. Select the metric to measure

(C) Loss performance of clustering.
(D) None of these (A) MSE
(B) MEA
33. If the accuracy of a non-
linear regression (robust to (C) Scatter
outliers) is to be evaluated, (D) Award/Punishment
then which is the metric is
the best? 38. Supervised learning, given:
(A) MSE
(B) MEA (A) desired output only
(C) Precision (B) Sequence of actions
(D) Recall (C) Training data + desired out-
put
34. Measures the percentage of
(D) Training data only
the actual positive class that
is correctly predicted. It is 39. The value of recall of a
the capture rate. model is 0.5, and it has true
(A) Recall positives=60, then what is
(B) Specificity the value of false nega-
tives?
(C) F1 Score
(A) 50
(D) none of above
(B) 55
35. What would make a robot
(C) 60
intelligent?
(D) 65
(A) It responds to the environment.
40. Expectation of predicted
(B) It responds to the environment target feature is low and
according to previous experiences. loss=0, then what will the
value of expectation of ac-
(C) It calculates mathematical tual target feature?
problems faster than human (A) Low
minds. (B) High
(D) It can jump 1.5 meters higher (C) Average
than humans.
(D) None of these
36. Every function has a
probability distribution func- 41. Suppose, you designed a
tion. model to predict chances of
a rain on a particular day.
(A) Continuous
Which metric will you use
(B) Discrete to measure accuracy of the
(C) Categorical model?
(D) Random (A) F-1 score

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
8 Chapter 1. Machine Learning

(B) Preceision (C) Multi-disciplinary

(C) Recall (D) All of the Above
(D) Mean square error
47. Which one of the statement
42. Which of the following is true regarding residuals

NARAYAN CHANGDER
methods do we use to best in regression analysis?
fit the data in Logistic Re- (A) Mean of residuals is always
gression? zero
(A) Least Square Error (B) Mean of residuals is always
(B) Any distance metric less than zero
(C) Maximum Likelihood (C) Mean of residuals is always
(D) Least Square Error and Maxi- greater than zero
mum Likelihood (D) There is no such rule for resid-
uals.
43. The most commonly used
cost function for linear re- 48. In the regression equation Y
gression is = a+bX, the Y is called:
(A) Bias (A) Independent variable
(B) RMSE
(B) Continuous variable
(C) Variance
(C) Dependent variable
(D) none of above
(D) None of the above
44. Determine the gradient of
the function x2 − 2y2 − 4y at 49. A regression model has ob-
point (0, 0)? tained MSE=0, then which of
the following is/are false?
(A) 2i-4j
(A) The model may be overfitted.
(B) 0i-4j
(B) The model may be underfitted.
(C) 0i+0j
(D) -4i-4j
(C) The model gives zero training
45. What is the relationship error.
between model and algo- (D) None of these
rithm?
(A) Algorithm + model = data 50. Performance metrics use for
Regression
(B) Algorithm = Model + data
(A) Mean squared error (MSE)
(C) Model = Algorithm(data)
(B) Confusion Matrix
(D) Data = Algorithm(model)
(C) Sensitivity, Specificity, Accu-
46. Machine learning is field racy
(A) Inter-disciplinary (D) Receiver Operating Charac-
(B) Single teristics (ROC) Curve

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.1 Introduction 9

51. In classification, output can (C) Deep Learning

only be (D) none of above
(A) 1
56. Which of the following eval-
(B) 2
uation metrics can not be ap-
(C) 3 plied in case of logistic re-
(D) 4 gression output to compare
with target?
52. Predicting the amount of
(A) AUC-ROC
rainfall in a region based on
various cues is a prob- (B) Accuracy
lem. (C) Mean-Squared-Error
(A) Supervised learning (D) None of these
(B) Unsupervised learning
57. The purpose of simple lin-
(C) Clustering
ear regression analysis is
(D) None of the above to:
53. Which of the following (A) Replace points on a scatter di-
yields discrete categorical agram by a straight-line
outputs? (B) Obtain the expected value of
(A) Anomaly detection the independent random variable
for a given value of the dependent
(B) Regression
variable
(C) Classification
(C) Predict one variable from an-
(D) Feature Learning other variable
54. The difference between the (D) none of above
average prediction of the
model and the predicted 58. Which of the below is not
value is called as supervised learning?
(A) Bias (A) Spam email identification
(B) Variance (B) House price prediction
(C) Loss (C) Speech recognition
(D) None of these (D) Amazon product suggestions
55. It is an interdisciplinary 59. What is meant by the term
field of scientific methods, ‘accuracy’?
processes, algorithms and
(A) The lack of bias in the data.
systems to extract knowl-
edge or insights from data (B) The level of detail at which
in various forms, either data is stored.
structured or unstructured. (C) The extent to which a value ap-
(A) Data Science proaches its true value.
(B) Machine Learning (D) The overall quality of the data.

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
10 Chapter 1. Machine Learning

60. Who invented the Percep- is labeled, which means that

tron which was a very, very each data tagged with the
simple classifier but when it correct label.
was combined in large num- (A) Supervised Learning
bers, in a network, it be-

NARAYAN CHANGDER
came a powerful monster? (B) Unsupervised Learning
(A) Frank Rosemarie (C) Reinforcement Learning
(B) Frank Rosenblet (D) Learning and Techniques
(C) Frank Rosenblatt 65. What doe ENIAC stands
(D) Frank Rosenbat for?
(A) Electric Number Intersect A
61. Which one in the following
Calculator
is not Machine Learning dis-
ciplines? (B) Electronic Numerical Integra-
tor and Computer
(A) Information Theory
(C) Engineering Numbering Into
(B) Neurostatistics
Auto Correct
(C) Optimization + Control
(D) Enter Number In Auto Correct
(D) Physics
66. The k-means algorithm is a
62. At what stage of the data
(A) Supervised learning algorithm
science lifecycle do you op-
timize the parameters?
(B) Unsupervised learning algo-
(A) Training the model
rithm
(B) Evaluating the model
(C) Semi-supervised learning al-
(C) Deploying the model gorithm
(D) none of above (D) Weakly supervised learning
algorithm
63. A regression based model
obtained the predicted 67. Is it possible to design a lin-
value of target feature as ear and logistic regression
2, however the actual value using neural network?
is 3. What is the value of
(A) Yes
MSE for the above given in-
stance? (B) No
(A) 0 (C) Cannot say
(B) 1 (D) May be possible
(C) 0.5 68. Who is the Chess grand
(D) None of these master beaten in a game by
I.B.M.’s system?
64. In this kind of Machine
Learning, an AI system is (A) Gary Kapov
presented with data which (B) Garry Kasper

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.1 Introduction 11

(C) Gary Kasparov (C) Same variance

(D) Gary Kerpov (D) None of the above

69. is a branch of ma- 74. Factors which affect perfor-

chine learning that uses mance of a learner system
data, loads and loads of does not include
data, to teach computers (A) Representation scheme used
how to do things only hu-
(B) Training scenario
mans were capable of be-
fore. (C) Type of feedback
(A) Supervised Learning (D) Good data structures
(B) Deep Learning 75. Classification is divided into
(C) Unsupervised Learning generative classifier and
(D) none of above (A) artificial intelligence
(B) deep learning
70. Which of the following is
not type of learning? (C) discriminative
(A) Semi-unsupervised Learning (D) probability
(B) Unsupervised Learning 76. What year did Artifical In-
(C) Supervised Learning telligence first stirred excite-
ment?
(D) Reinforcement Learning
(A) 1905
71. Which model is the best (B) 1950
choice for fraud detection?
(C) 1590
(A) Precision = 0.8, Recall = 0.4
(D) 1915
(B) Precision = 0.4, Recall = 0.8
77. What year was the
(C) Precision = 0.7, Recall = 0.7
E.N.I.A.C. first invented?
(D) none of above
(A) 1904
72. Why do we need to set (B) 1914
hyper-parameters?
(C) 1940
(A) Model configuration
(D) 1490
(B) Model validation
78. is a simple approach to
(C) Model deployment
supervised learning. It as-
(D) none of above sumes that the dependence
of Y on X1, X2, . . . Xp is
73. As the number of training
linear.
examples goes to infinity,
your model trained on that (A) Linear regression
data will have: (B) Logistic regression
(A) Lower variance (C) Gradient Descent
(B) Higher variance (D) All of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
12 Chapter 1. Machine Learning

79. I.B.M.’s system beat the (B) Predict a continuous variable

world champion of chess, from discrete variables.
the grand-master. (C) Predict a continuous variable
(A) Deep Sea from continuous or discrete vari-
(B) Deep Ocean ables.

NARAYAN CHANGDER
(C) Deep Lake (D) None of these
(D) Deep Blue
84. To predict which customers
80. algorithms enable the are more likely to purchase
computers to learn from BANCA, a machine learning
data, and even improve needs historical data of cus-
themselves, without being tomers who:
explicitly programmed. (A) bought banca
(A) Artificial Intelligence (B) declined to buy banca
(B) Machine Learning
(C) both bought banca and who
(C) Deep Learning declined to buy banca
(D) Traditional Learning (D) all customers of TCB
81. The model learns and up-
85. Unsupervised Learning in
dates itself through re-
Machine learning means
ward/punishment in case
of (A) no labelled data
(A) Supervised learning algorithm (B) labeled data
(C) great algorithm
(B) Unsupervised learning algo-
(D) none of the above
rithm
(C) Semi-supervised learning al- 86. What year did Machine
gorithm Learning began to flourish?
(D) Reinforcement learning algo- (A) 1980
rithm
(B) 1918
82. The model effectiveness is (C) 1908
measured using
(D) 1890
(A) Bias
(B) Variance 87. The value of precision of a
model is 0.5, and it has true
(C) Loss
positives=60, then what is
(D) All of these the value of false positives?
83. Logistic regression is used (A) 50
when you want to:
(B) 55
(A) Predict a discrete variable
from continuous or discrete vari- (C) 60
ables. (D) 65

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 1.2 Machine Learning and applications 13

88. Logistic regression assumes (B) Learning from unlabeled data

that
(A) The independent variables (C) Learning from environment
should be independent of each
other. That is, the model should (D) none of above
have little or no multicollinearity.
90. ML is a field of AI consist-
(B) The independent variables are
ing of learning algorithms
linearly related.
that?
(C) Both of above
(A) Improve their performance
(D) None of these
(B) At executing some task
89. What is the concept of rein-
forcement learning? (C) Over time with experience
(A) Learning from labeled data (D) All of the above

1.2 Machine Learning and applications

1. How do you choose the root Which of the following func-
node while constructing a tion is used by logistic re-
Decision Tree? gression to convert the prob-
ability in the range between
(A) An attribute having high en-
[0, 1].
tropy
(A) Sigmoid
(B) An attribute having largest in-
formation gain (B) Mode
(C) An attribute having high en- (C) Square
tropy and Information gain (D) All of the above
(D) None of the Mentioned
4. If the null hypothesis is false
2. The residual is the difference then which of the following
in between is accepted?

(A) actual value of y and the esti- (A) Null Hypothesis

mated value of y (B) Positive Hypothesis
(B) actual value of x and the esti- (C) Negative Hypothesis
mated value of x
(D) Alternative Hypothesis
(C) actual value of y and the esti-
mated value of x 5. Weather forecasting is an
example of
(D) actual value of x and the esti-
mated value of y (A) Clustering
(B) Classification
3. Logistic Regression trans-
forms the output probability (C) Regression
to be in a range of [0, 1]. (D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
14 Chapter 1. Machine Learning

6. In a Decision Tree Leaf Node (C) Gaussian classifier

represents (D) Maximum margin classifier
(A) One of the Class Label
11. Which of the following mea-
(B) One of the complete observa- sures best analyse the per-

NARAYAN CHANGDER
tion formance of a classifier?
(C) One of the attribute (A) Precision
(D) None of the Mentioned (B) Recall
(C) Accuracy
7. KNN is algorithm
(D) Depends on the application
(A) Non-parametric and Lazy
Learning 12. KNN algorithm requires
(B) Parametric and Lazy Learning (A) More time for training
(C) Parametric and Eager Learn- (B) More time for testing
ing (C) Equal time for training and
(D) Non-parametric and Eager testing
Learning (D) None of the Mentioned

8. Considering the K-median 13. Starting from own cluster

algorithm, if points (0, 3), (2, and moving towards a sin-
1), and (-2, 2) are the only gle cluster is termed as
points which are assigned to (A) Agglomerative clustering
the first cluster now, what (B) Divisive clustering
is the new centroid for this
(C) K-means
cluster?
(D) K-mediods
(A) (0, 2)
(B) (2, 1) 14. Agglomerative clustering
uses method
(C) (2, 0)
(A) Bottom up
(D) (1, 2)
(B) Top down
9. If 20 is added to the data set (C) Greedy
below, which statement will (D) None
be true?50, 55, 55, 60, 62
15. Which among the following
(A) The median will be 20
is not true for a Bayesian
(B) The mode will increase classifier?
(C) The mean will increase (A) A (natural) class is to predict
(D) The mean will decrease the values of features for members
of that class.
10. Maximum aposteriori clas- (B) It is a probabilistic model.
sifier is also known as:
(C) It is based on Bayes’ theorem.
(A) Decision tree classifier
(B) Bayes classifier (D) It is not used in data mining.

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.2 Machine Learning and applications 15

16. Let A be an example, and C (D) All of the above

be a class. The probability
P(C) is known as: 21. Text Recognition (spam or
no-spam) is an example of
(A) Apriori probability
(A) Classification
(B) Aposteriori probability
(B) Clustering
(C) Class conditional probability
(C) Regression
(D) None of the above
(D) none of above
17. What is the purpose of re-
stricting hypothesis space in 22. Non-leaf nodes of decision
machine learning? tree correspond to:
(A) Can be easier to search (A) Attributes
(B) May avoid overfit since they (B) Classes
are usually simpler (e.g. linear or (C) Data instances
low order decision surface)
(D) Node of the above
(C) Both of the above
(D) None of the above 23. Regression is used in:
(A) Prediction
18. Divisive clustering uses
method (B) Exploration
(A) Bottom up (C) Descriptive data mining
(B) Top down (D) Explanative data mining
(C) Greedy 24. Which of the following
(D) None statements is true about lin-
ear regression regarding
19. Which of the following outlier?
methods do we use, to find
the best fit line for data in (A) Linear regression is sensitive
Linear Regression? to outlier
(A) Least Square Error (B) Linear regression is not sensi-
tive to outliers
(B) Maximum Likelihood
(C) Depends upon the data
(C) Logarithmic Loss
(D) None of the above
(D) Both A and B
25. What will happen when
20. Which of the following ne-
eigenvalues are roughly
cessitates feature reduction
equal in PCA?
in machine learning?
(A) PCA will perform outstand-
(A) Irrelevant and redundant fea-
ingly
tures
(B) Limited training data (B) PCA will perform badly

(C) Limited computational re- (C) Can’t Say

sources (D) None of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
16 Chapter 1. Machine Learning

26. Which of the following is re- 31. Decision tree is an algo-

quired by K-means cluster- rithm for:
ing? (A) Classification
(A) defined distance metric (B) Clustering
(B) number of clusters

NARAYAN CHANGDER
(C) Association rule mining
(C) initial guess as to cluster cen- (D) Noise filtering
troids
32. State whether the state-
(D) all answers are correct
ments are True or False.Statement
27. Machine Learning is a A:When the hypothesis
branch of.. space is richer, overfitting
is more likely.Statement
(A) AI B:When the feature space
(B) Java is larger, overfitting is more
(C) c likely.
(D) c++ (A) False, False
(B) True, False
28. Which of the following is an
(C) True, True
unsupervised task?
(D) False, True
(A) Learning to play chess
(B) Predicting if an edible item is 33. If the given dataset contains
sweet based on the information of 100 observations out of 50
the ingredients and their quanti- belongs to class1 and other
ties 50 belongs to class2. What
will be the entropy of the
(C) Grouping related documents
given dataset?
from an unannotated corpus
(A) 0
(D) All of the above
(B) 1
29. The most commonly used (C) -1
measure of similarity is the
or its square. (D) 0.5

(A) euclidean distance 34. Consider two binary at-

tribute X and Y. We know
(B) city-block distance
that the attribute are in-
(C) Chebychev’s distance dependent and probability
(D) Manhattan distance P(X=1) = 0.6, and P(Y=0) =
0.4. What is the probability
30. Estimating the price of a that both X and Y have val-
house is an example of ues 1?
(A) Clustering (A) 0.36
(B) Classification (B) 0.4
(C) Regression (C) 0.6
(D) none of above (D) 0.76

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.2 Machine Learning and applications 17

35. The entropy of a given (A) Input value and output value
dataset is zero. This state-
(B) Input and target value
ment implies what?
(C) Output value and target value
(A) further splitting is required
(B) no further splitting is required
(D) Model parameter and output
(C) Need some other information value
to decide splitting
(D) None of the Mentioned 40. Decision trees can be used
for:
36. What type of Machine
Learning Algorithm is suit- (A) Classification only
able for predicting the de- (B) Regression only
pendent variable with two
(C) Both Classification and Re-
different values?
gression
(A) Logistic Regression
(D) Neither of Classification and
(B) Linear Regression Regression
(C) Multiple Linear Regression
41. Leaf nodes of a decision
(D) Polynomial Regression
tree correspond to:
37. The goal of clustering a set (A) Attributes
of data is to
(B) Classes
(A) divide them into groups of
data that are near each other (C) Data instances
(B) choose the best data from the (D) Node of the above
set
42. Data used to build a model
(C) determine the nearest neigh-
in machine learning model.
bors of each of the data
(D) predict the class of data (A) Validation data
(B) Training data
38. If we decrease the input
variable by one unit in a sim- (C) Test data
ple linear regression model. (D) Hidden data
How many units of the out-
put variable will change? 43. clustering seeks the min-
(A) reduced by Intercept imum distance between any
records in two clusters to
(B) increased by Intercept
find the distance between
(C) increased by Slope the points in two clusters.
(D) reduced by Slope (A) Single linkage
39. Regression finds out the (B) Complete Linkage
model parameter which pro-
(C) Average Linkage
duces the least square error
between: (D) None

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
18 Chapter 1. Machine Learning

44. The Number of coefficients (B) When irrelevant attributes

required to estimate a sim- have been removed from the data.
ple linear regression?
(A) 1 (C) When a generalized model of
the data is desirable

NARAYAN CHANGDER
(B) 2
(C) 0 (D) When an explanation of what
has been found is of primary im-
(D) 3 portance.
45. K-fold cross-validation is 50. Cluster analysis is
(A) linear in K (A) Unsupervised Learning
(B) quadratic in K (B) Supervised learning
(C) cubic in K (C) Semi supervised learning
(D) exponential in K (D) None of these
46. Find the outliers in the given 51. Which of the following
data set below. 27, 20, 1, 3, algorithm is called Lazy
30, 31, 33 Learner?
(A) 1, 3 (A) KNN
(B) 27, 33 (B) SVM
(C) 31, 33 (C) Naive Bayes
(D) 1, 33 (D) Decision Tree
47. The output of regression al- 52. PCA works better if there
gorithm is usually a: is?(i) A linear structure in the
(A) Real variable data(ii) If the data lies on a
curved surface and not on
(B) Integer variable
a flat surface(iii) If variables
(C) Character variable are scaled in the same unit
(D) String variable (A) 1 and 2

48. Estimating the price of a (B) 2 and 3

smartphone is an example (C) 1 and 3
of (D) 1, 2 and 3
(A) Clustering
53. A statement made about a
(B) Classification population for testing pur-
(C) Regression pose is called?
(D) none of above (A) Statistic

49. A nearest neighbour ap- (B) Hypothesis

proach is best used (C) Level of Significance
(A) With large-sized datasets. (D) Test-Statistic

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 1.3 Types of Machine Learning 19

54. You are given reviews of (A) Supervised Learning

movies marked as positive, (B) Unsupervised Learning
negative, and neutral. Clas-
sifying reviews of a new (C) Reinforcement Learning
movie is an example of (D) None of these

1.3 Types of Machine Learning

1. One of the types of prob- 4. It is a type of machine learn-

lems solved using machine ing where an agent is put
learning where data points in an environment and he
are assign into clusters. The learns to behave in this envi-
main aim is to group similar ronment by performing cer-
items. tain actions and observing
(A) Regression the rewards which it gets
from those actions.
(B) Classification
(A) Supervised Learning
(C) Clustering
(B) Unsupervised Learning
(D) none of above (C) Reinforcement Learning
2. It is a type of machine (D) none of above
learning where it involves
a guide. It is also a tech- 5. One of the types of prob-
nique in which we teach or lems solved using machine
train the machine using data learning where the output is
which is well labelled. a categorical quantity and
the main aim is to compute
(A) Supervised Learning the category of the data.
(B) Unsupervised Learning (A) Regression
(C) Reinforcement Learning (B) Classification
(D) none of above (C) Clustering
(D) none of above
3. One of the types of prob-
lems solved using machine 6. It is the training of machine
learning where the output using information that is un-
variable is a continuous labeled and allowing the al-
quantity and the main aim gorithm to act on that infor-
is to forecast or predict. mation without guidance.
(A) Regression (A) Supervised Learning
(B) Classification (B) Unsupervised Learning
(C) Clustering (C) Reinforcement Learning
(D) none of above (D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 20 Chapter 1. Machine Learning

1.4 Supervised Learning

1. What kind of learning al- 5. What kind of learning algo-
gorithm for “Future stock rithm for “Facial identities or
prices or currency exchange facial expressions”?

NARAYAN CHANGDER
rates”? (A) Recognizing Anomalies
(A) Recognizing Anomalies (B) Prediction
(B) Prediction (C) Generating Patterns
(C) Generating Patterns (D) Recognition Patterns
(D) Recognition Patterns 6. The most widely used met-
rics and tools to assess a
2. Fraud Detection, Image Clas-
classification model are:
sification, Diagnostic, and
Customer Retention are ap- (A) Confusion matrix
plications in (B) Cost-sensitive accuracy
(A) Unsupervised Learning:Clustering (C) Area under the ROC curve
(D) All of them
(B) Supervised Learning:Classification
7. Which of these can be eval-
uated by a confusion ma-
(C) Reinforcement Learning trix?
(D) Unsupervised Learning:Regression (A) Recall
(B) F1 Score
3. is the opposite of unsu-
pervised learning (C) Precision
(A) unsupervised (D) ALL
(B) AI element 8. Which of the following is
(C) clustering an example of supervised
learning
(D) Supervised
(A) Nearest Neighbour
4. Supervised learning is when (B) Regression
you give a machine an input (C) classification
and it gives you an output
(D) Hi

(A) Depending on the environ- 9. A field in the dataset used in

ment the machine learning algo-
rithm. E.g. happy and sad
(B) Depending on the input you
gave it (A) dataset
(C) Magic (B) supervised learning

(D) Depending on what the AI (C) unsupervised learning

agent like (D) classifiers

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
1.4 Supervised Learning 21

10. Supervised Learning is: (C) Predictive models

(A) learning with the help of ex- (D) Testing
amples
15. The technique of adjusting
(B) learning without teacher
different values for the pa-
(C) learning with the help of rameters of a classification
teacher method is termed as
(D) learning with computers as su- (A) Feature extraction
pervisor
(B) Hyper parameter tuning
11. Supervised learning is an (C) Data pre-processing
system
(D) Data modelling
(A) Artificial intelligence
(B) Machine learning 16. In supervised learning:
(C) Computer science (A) classes are not predefined
(D) Deep learning (B) classes are predefined

12. Which of the following is (C) classes are not required

a widely used and effective (D) classification is not done
machine learning algorithm
based on the idea of bag- 17. Suppose you are working
ging? on stock market prediction,
and you would like to pre-
(A) Decision Tree
dict the price of a particular
(B) Regression stock tomorrow (measured
(C) Classification in dollars). You want to
use a learning algorithm for
(D) Random Forest
this.
13. Which of the following is (A) Classification
NOT supervised learning?
(B) Regression
(A) PCA
(C) Clustering
(B) Decision Tree
(D) none of above
(C) Linear Regression
(D) SVM 18. You are given data about
seismic activity in Japan and
14. After training the ML you want to predict a magni-
model, we see how ac- tude of the next earthquake.
curately it predicts the This is an example of
answer/responds. For
(A) Supervised Learning
example-does it cry when I
say something mean to it? (B) Unsupervised learning
(A) Recognition (C) Serration
(B) Training (D) Dimensionality reduction

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
NARAYAN CHANGDER
 2. Application of supervised learning

1. Which of the following op- 4. A machine learning model

tions is true about KNN al- uses labelled data to predict
gorithm? whether a customer will pur-
(A) It can be used for classification chase a product or not. This
modelling technique comes
under
(B) It can be used for regression
(A) Regression
(C) It can be used in both classifi-
(B) Classification
cation and regression
(C) Clustering
(D) None of these
(D) None of these
2. Which of the following al-
gorithms is not an example 5. The SVM’s are less effective
of ensemble learning algo- when:
rithm? (A) The data is linearly separable.
(A) Random Forest
(B) The data is clean and ready to
(B) Adaboost
use.
(C) Decision Trees
(C) The data contains overlap-
(D) Gradient Boosting ping points.
3. KNN is a algorithm. (D) None of these

(A) Non parametric and lazy 6. In case of classification prob-

learning lems, Random Forest uses
for its final prediction.
(B) Parametric and lazy learning
(A) Minority vote
(C) Parametric and eager learn-
ing (B) Majority vote
(D) Non parametric and eager (C) Sum of all predictions
learning (D) Average of all predictions

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 24 Chapter 2. Application of supervised learning

2.1 Unsupervised learning

1. In unsupervised learning, 6. The most accurate clustering
the machine learns by model is

NARAYAN CHANGDER
(A) Identifying patterns in the (A) Agglomerative HC
dataset
(B) Divisive HC
(B) Using class labels
(C) k-Means
(C) None of these
(D) Both of these (D) Depends on dataset

2. In unsupervised learning, 7. What is the topic that i talk

the dataset will be about?
(A) Labelled (A) Unsupervised learning
(B) Semi-labelled
(B) Supervised learning
(C) None of these
(C) Over fitting
(D) Non-labelled
(D) AI element
3. What does GMM-EM opti-
mise? 8. is not a model in cluster-
(A) Minimises the average dis- ing
tance between the samples and (A) Agglomerative HC
the mean of the nearest Gaussian
(B) Divisive HC
(B) Minimises the negative-log- (C) k-Means
likelihood of the model
(D) Random Forest
(C) Maximises the classification
rate 9. Optimum number of clus-
(D) none of above ters in agglomerative HC is
found out using
4. Types of clustering include
(A) Dendrogram
(A) Hierarchical clustering
(B) Elbow Method
(B) k-NN
(C) Both of these (C) Random State
(D) None of these (D) None of these

5. is an unsupervised learn- 10. What is my name:

ing type
(A) Nasser
(A) Association
(B) Clustering (B) Nasser Saeed

(C) Both of these (C) Mohammed

(D) None of these (D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.2 Supervised and unsupervised learning 25

11. Is it possible that the as- 14. Optimum number of clus-

signment of observations to ters in k-Means is found out
clusters doesn’t change be- using
tween successive iterations
(A) Dendrogram
in K-Means?
(B) Elbow Method
(A) yes
(B) no (C) Random State
(C) can’t say (D) None of these
(D) none of above 15. Hierarchical clustering is
12. Agglomerative hierarchical classified into
clustering is also known as (A) 3
(A) Top down approach (B) 4
(B) Bottom up approach
(C) 2
(C) Both of these
(D) None of these
(D) None of these
16. What is unsupervised learn-
13. Example of unsupervised
ing
learning
(A) Nothing (A) is the training of an artificial in-
telligence (AI) algorithm using in-
(B) discover formation that is neither classified
(C) Clustering nor labeled and allowing the al-
(D) classified into two categories gorithm to act on that information
of algorithms:Clustering:A cluster- without guidance.
ing problem is where you want to (B) Probabilistic
discover the inherent groupings in
(C) clustering
the data, such as grouping cus-
tomers by purchasing behavior. (D) none of above

2.2 Supervised and unsupervised learning

1. Which if these is not a pa- (A) Representation Learning
rameter for multi class neu-
(B) Classification
ral network?
(C) Regression
(A) Regularization Weight
(D) none of above
(B) Number of Hidden Nodes
(C) Learning Rate 3. The following descriptions
(D) Number of Learning Iterations best describe what:1. Value
that has to be assigned man-
2. Which of these is not a ually. 2. The K value in
supervised learning algo- K-nearest-neighbor is an ex-
rithm? ample of this. 3. Value is set

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 26 Chapter 2. Application of supervised learning

before the training. (D) Unsupervised learning

(A) Centroid 6. Select the odd one out.
(B) Model Parameter (A) AUC
(C) Model hyper parameter (B) ROC

NARAYAN CHANGDER
(D) Significance (C) Logistic Regression
4. A value that defines the step (D) none of above
taken at each iteration, be-
fore correction? 7. Logistic regression is?

(A) Gradient descent (A) Supervised machine learning

algorithm
(B) learning rate
(B) Unsupervised machine learn-
(C) l2 regularization ing algorithm
(D) l1 regularization (C) Both
5. is a learning model that (D) none of above
is used to identify a re-
lationship between large 8. Which of these isn’t a classi-
amounts of information fication type?
from a data set. (A) Binary classification
(A) Classification (B) Multiclass multilabel
(B) Multi class classification (C) Singleclass multilabel
(C) Association (D) Multiclass Single Label

2.3 Classification in Machine learning

1. What will happen when (C) it has to formed from a binary
eigenvalues are roughly tree
equal?
(D) The output can be easily un-
(A) PCA will perform outstand- derstood by a layman
ingly
3. How do you handle miss-
(B) PCA will perform badly ing or corrupted data in a
(C) Can’t Say dataset?
(D) None of above (A) Drop missing rows or columns

2. Which is FALSE about Deci- (B) imputation can be done

sion Tree
(C) Assign a unique category to
(A) We can extract RULES from the missing values
Decision Tree
(D) All of the above
(B) It is good to find relationship
of non-linearly correlated data 4. Perceptron is

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.3 Classification in Machine learning 27

(A) linear classifier 9. Simple regression is a

relationship between 2 or
(B) non Linear Classifier
more variables
(C) none of the above
(A) Linear
(D) none of above (B) Non Linear
5. What is pca components (C) Categorical

(A) Set of all eigen vectors for the (D) Systematical

projection space 10. What is the purpose of per-
(B) Matrix of principal compo- forming cross-validation?
nents (A) To assess the predictive perfor-
(C) Result of the multiplication mance of the models
matrix (B) To judge how the trained
(D) None of the above options model performs outside the sam-
ple on test data
6. The technique Which is used (C) Both A and B
to re-scales features with a
(D) none of the above
distribution value between
0 and 1 11. Which of the following sen-
(A) Normalization tence is FALSE regarding re-
gression?
(B) standization
(A) It relates inputs to outputs
(C) none of the above
(B) It is used for prediction
(D) both a& b (C) It may be used for interpreta-
tion.
7. Which of the following is not
a type of supervised learn- (D) It discovers causal relation-
ing? ships.

(A) Classification 12. Machine Learning has vari-

(B) Regression ous function representation,
which of the following is not
(C) Clustering numerical functions?
(D) None (A) Linear Regression
(B) Support Vector Machines
8. Which of this is based on the
probabilities of an event to (C) Neural Network
happen? (D) Case-based
(A) Bayesian network 13. What type of Machine
(B) Naive Bayes Learning Algorithm is suit-
able for predicting the con-
(C) Decision tree
tinuous dependent vari-
(D) SVM able?

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
28 Chapter 2. Application of supervised learning

(A) Logistic Regression 18. Which is true about Ran-

dom Forest (RF)
(B) Linear Regression
(A) RF completely a different ma-
(C) Decision Tree Classifier
chine learning technique
(D) KNN Classifier (B) RF combines multiple decision

NARAYAN CHANGDER
trees to make a decision
14. You are analyzing a scatter
plot that has a negative cor- (C) RF cannot handle continuous
relation. What is true for the data
x and y values? (D) none of above
(A) As x increases, y decreases 19. Appropriate chart for visu-
(B) As x increases, y increases alizing the linear relation-
ship between two variables
(C) As x decreases, y decreases
is
(D) none of above (A) Scatter plot
15. KNN algorithm appropriate (B) Barchart
for (C) Histograms
(A) Lower number of features (D) None of Mentioned
(B) Large number of features 20. All data is labeled and the
(C) No such restriction on number algorithms learn to predict
of features the output from the input
data
(D) None of the Mentioned
(A) Dataset
16. Which of this is the laziest (B) Classifiers
technique of learning?
(C) supervised learning
(A) k-nearest neighbour (D) unsupervised learning
(B) k-means
21. Which of the following is an
(C) k-fold example of a deterministic
(D) k-19 algorithm?
(A) PCA
17. What is the use of Vali-
(B) K-Means
dation dataset in Machine
Learning? (C) None of the above
(A) To train the machine learning (D) none of above
model. 22. KNN Algorithm can be used
(B) To evaluate the performance for
of the machine learning model (A) Only for Classification
(C) To tune the hyperparameters (B) Only for Regression
of the machine learning model
(C) Both Classification and Re-
(D) None of the above. gression

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.4 Support vector machine (SVM) 29

(D) None of the Mentioned (D) none of the above

23. Which of the following is a 24. * if the model/algorithm
good test dataset character- shows low bias but high
istic? variance then it is
(A) Large enough to yield mean- (A) overfitting
ingful results
(B) underfitting
(B) Is representative of the dataset
as a whole (C) goodness of fit
(C) Both A and B (D) none of the above

2.4 Support vector machine (SVM)

1. What do you mean by a 4. Consider these are the out-
hard margin? put values (event happened)
(A) The SVM allows very low error { 10, 20, 35, 45} . the Pre-
in classification diction Line is 10 and max-
imal margin is +5 and-5.
(B) The SVM allows high amount Apply the SVM regression
of error in classification method to find the largest
(C) None of the above error rate.
(D) none of above (A) 5

2. Which of the following are (B) 30

real-world applications of (C) 15
the SVM?
(D) 25
(A) Text and Hypertext Categoriza-
tion 5. Consider the following
(B) Image Classification points w=10, input X is 5
and the Constant value is
(C) Clustering of News Articles 10, what is the predictor
(D) All of the above value of Y in linear regres-
sion method
3. What do you mean by gen-
(A) 100
eralization error in terms of
the SVM? (B) 200
(A) How far the hyperplane is (C) 60
from the support vectors
(D) 80
(B) How accurately the SVM can
predict outcomes for unseen data 6. Support Vector Machine,
How to reduce the error rate
than Linear regression .
(C) The threshold amount of error
in an SVM (A) Hyperplane
(D) none of above (B) Associative

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
30 Chapter 2. Application of supervised learning

(C) Clustering points 11. Which of the following algo-

(D) Distance rithms do we use for Vari-
able Selection?
7. Suppose you are dealing
with 4 class classification (A) LASSO

NARAYAN CHANGDER
problem and you want to (B) Ridge
train a SVM model on the
data for that you are us- (C) Both
ing One-vs-all method. How
(D) None of these
many times we need to
train our SVM model in such
case? 12. A term used to describe
the case when the indepen-
(A) 1
dent variables in a multiple
(B) 2 regression modelare corre-
(C) 3 lated is
(D) 4 (A) Regression
8. What would happen when (B) correlation
you use very small C (C 0)?
(C) multicollinearity
(A) Misclassification would hap-
pen (D) none of the above
(B) Data will be correctly classified
13. Sentiment Analysis is an
(C) Can’t say example of:a)Regression,
b)Classificationc)Clusteringd)Reinforcemen
(D) None of these
Learning
9. The effectiveness of an SVM
(A) a only
depends upon
(A) Selection of Kernel (B) a and b
(B) Kernel Parameters (C) a and c
(C) Soft Margin Parameter C
(D) a, b and d
(D) All of the above
10. What is/are true about ker- 14. What are support vector
nel in SVM?1. Kernel func- machines used for?
tion map low dimensional (A) classification, outliers detec-
data to high dimensional tion, regression
space2. It’s a similarity func-
tion (B) regression
(A) 1 (C) outliers detection, classifica-
(B) 2 tion
(C) 1 and 2 (D) clustering, classification, out-
(D) None of these liers detection, regression

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.5 Reinforcement Learning 31

15. When the C parameter is set (D) none of above

to infinite, which of the fol-
lowing holds true? 16. Logistic regression is a
regression technique that is
(A) The optimal hyperplane if ex- used to model data having
ists, will be the one that completely a outcome.
separates the data (A) linear, numeric
(B) The soft-margin classifier will (B) linear, binary
separate the data (C) nonlinear, numeric
(C) None of the above (D) nonlinear, binary

2.5 Reinforcement Learning

1. A computer or robot that (A) the computer choose the one
can understand the envi- with no reward
ronment using sensors and (B) the computer choose the one
take actions to deal with with the lowest reward
that environment.
(C) the computer choose the one
(A) DFS with the biggest reward
(B) AI Agent (D) none of above
(C) BFS
3. Reinforcement learning is a
(D) none of above type of learning that uses
2. Tree search is used in re- (A) reward and punishment
inforcement learning algo- (B) labels and features
rithms to build tree of all
possible paths to reach a tar- (C) only features
get. so (D) none of above

2.6 Linear Regression

1. What category of machine (A) r2
learning algorithm finds pat- (B) r
terns in the data when the
data is not labeled? (C) y hat
(A) Supervised learning (D) x
(B) Unsupervised learning 3. To test linear relation-
(C) Reinforcement learning ship of y(dependent) and
x(independent) continuous
(D) All variables, which of the fol-
2. The coefficient of determina- lowing plot best suited?
tion is (A) Scatter Plot

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
32 Chapter 2. Application of supervised learning

(B) Bar Chart 8. Which of the following state-

(C) Histogram ment is true about outliers in
Linear regression?
(D) None of these
(A) Linear regression is sensitive
4. A multiple regression model to outliers

NARAYAN CHANGDER
has the form:y = 8-3 X1 + 4 (B) Linear regression is not sensi-
X2 .As X1 increases by 1 unit tive to outliers
(holding X2 constant), y will
(C) Can’t say
(A) decrease by 4 units (D) None of these
(B) increase by 4 units
9. Simple linear regression
(C) decrease by 3 units (SLR) is a type of statis-
(D) increase by 3 units tical analysis.
(A) Univariate
5. FALSE
(B) Bivariate
(A) Least Square Error
(C) Multivariate
(B) Maximum Likelihood
(D) none of above
(C) Logarithmic Loss
(D) Both A and B 10. Which of the following eval-
uation metrics can be used
6. Using Ridge, What will hap- to evaluate a model while
pen when you apply very modeling a continuous out-
large penalty? put variable?
(A) Some of the coefficient will be- (A) AUC-ROC
come absolute zero
(B) Accuracy
(B) Some of the coefficient will ap-
proach zero but not absolute zero (C) Logloss
(D) Mean-Squared-Error
(C) Both A and B depending on
the situation 11. If I tell you that the correla-
tion between two variables
(D) None of these
is .50, what is the variance
7. Suppose you have to pre- accounted for (i.e. what %
dict the salary of employees of the change in y is ex-
from their experience. This plained by the change in
is: x)?

(A) Classification task (A) .50

(B) Regression task (B) .25
(C) Clustering task (C) .75
(D) None (D) .10

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.6 Linear Regression 33

12. Suppose that in our sam- 16. In MLR, the square of the
ple, following a logistic re- multiple correlation coeffi-
gression analysis, the odds cient or R2 is called the
for girls of having a posi- (A) Variance
tive attitude to school were
1.25. What proportion of (B) Covariance
girls would be expected to (C) Coefficient of determination
have a positive attitude?
(D) Big R
(A) 25%
17. A shows the relation-
(B) 35% ship between two quantita-
(C) 45% tive variables measured on
the same individuals.
(D) 55%
(A) scatterplot
13. Which of the following is (B) residual plot
true about Residuals?
(C) regression line
(A) Lower is better
(D) television
(B) Higher is better
18. A regression model in
(C) A or B depend on the situation which more than one inde-
pendent variable is used to
(D) None of these predict the dependent vari-
able is called as
14. A term used to describe the
(A) a simple linear regression
case when the independent
model
variables in a multiple re-
gression model are corre- (B) a multiple regression model
lated is (C) an independent model
(A) independence of residual (D) none of above
(B) linearity of residual 19. The measures the
(C) multicollinearity strength of the linear rela-
tionship between two quan-
(D) homocedasticity
titative variables.
15. What type of machine (A) correlation
learning algorithm makes (B) regression line
predictions when you have
a set of input data with la- (C) residual
bels? (D) scatterplot
(A) Supervised learning 20. The equation of the least
(B) Unsupervised learning squares regression line is y
(hat) = 1.3 + 0.73x. What is
(C) Reinforcement learning
the residual for the point (4,
(D) All 7)?

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
34 Chapter 2. Application of supervised learning

(A) 2.78 (D) A person’s age and what kind

(B) 3.00 of cereal they eat.

(C) 4.00 25. A linear regression (LR)

(D) 4.22 analysis produces the equa-

NARAYAN CHANGDER
tion Y = 3 + 0.4X. This indi-
21. Residuals refer to the er- cates that:
ror values of the model.
Which of the following is (A) When Y = 0.4, X = 3
true about residuals? (B) When Y = 0, X = 3
(A) Lower is better (C) When X = 3, Y = 0.4
(B) Higher is better (D) When X = 0, Y = 3
(C) depend on situation
26. Which of the following
(D) None of the above
shows the BEST result of in-
22. Suppose you have been dependence of residual?
given a fair coin and you (A) 2.09
want to find out the odds of
getting heads. Which of the (B) 2.89
following option is true for (C) 1.89
such a case?
(D) 2.01
(A) odds will be 0
(B) odds will be 0.5 27. Stepwise regression method
allows researchers to cre-
(C) odds will be 1
ate the most parsimonious
(D) None model.
23. Which of these CANNOT be (A) Yes
described by a scatterplot?
(B) No
(A) direction
(C) Somewhat true
(B) strength
(D) Somewhat false
(C) form
(D) causation 28. What will happen when
you apply very large
24. Which example shows penalty in case of Lasso?
CAUSATION? (Which one
CAUSED the other to hap- (A) Some of the coefficient will be-
pen?) come zero

(A) High social media usage and (B) Some of the coefficient will be
reduced grades. approaching to zero but not abso-
lute zero
(B) Car ran out of gas and being
stranded on the side of the road. (C) Both A and B depending on
the situation
(C) Recess time and number of
friends. (D) None of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.7 Gradient Descent 35

29. Which of the following indi- 0.0001

cates a fairly strong relation-
ship between X and Y? (C) The t-statistic for the null hy-
pothesis Beta coefficient=0 is 30
(A) Correlation coefficient = 0.9
(B) The p-value for the null hy-
pothesis Beta coefficient =0 is (D) None of these

2.7 Gradient Descent

1. Which of the following is not (B) ADAGRAD
the a variant of Gradient De-
scent (C) ADAM
(A) Vanilla (D) None of the above

2.8 NEURAL NETWORK

1. Deep learning algorithm for (C) Softmax
sequential problem (D) tanh
(A) CNN
5. How many output layers
(B) ANN are required for construct-
(C) DBN ing an Artificial Neural Net-
(D) RNN work?
(A) 1
2. When the cell is said to be
fired? (B) 3
(A) if potential of body reaches a (C) 2
steady threshold values (D) none of above
(B) if there is impulse reaction
6. A consists of more than
(C) during upbeat of heart one perceptron
(D) none of the mentioned (A) multilayer perceptron
3. What is purpose of Axon? (B) single layer perceptron
(A) receptors (C) two layer perceptron
(B) transmitter (D) none of above
(C) transmission 7. type of model, the algo-
(D) none of the mentioned rithm learns from a dataset
which is labelled, and the
4. The most suitable activation algorithm uses the answer
function for hidden layer keys to evaluate its accu-
(A) Sigmoid racy on the training data.
(B) ReLu (A) Supervised learning

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
36 Chapter 2. Application of supervised learning

(B) UnSupervised learning 13. which function decides if

(C) Reinforcement learning the neuron should be acti-
vated or not.
(D) none of above
(A) Bias function
8. is a subset of Artificial In- (B) Activation function

NARAYAN CHANGDER
telligence.
(C) Trigger function
(A) DL
(D) none of above
(B) CV
(C) AI 14. Signal transmission at
synapse is a?
(D) none of above
(A) physical process
9. Which of the following is not
(B) chemical process
an AI application?
(C) physical & chemical both
(A) Google Maps
(D) none of the mentioned
(B) Writing suggestion in gmail
(C) Face Recognition 15. Which of the following is
an application of NN (Neu-
(D) Kahoot
ral Network)?
10. The part of the neuron that (A) Sales forecasting
takes information away
(B) Data validation
from the cell body is called
a(n) (C) Risk management
(A) dendrite (D) All of the mentioned
(B) axon 16. Subset of machine learning
(C) cell body in which artificial neural net-
works adapt and learn from
(D) Schwann cell
vast amount of data
11. When both inputs are 1, (A) Deep Learning
what will be the output of
(B) Data Science
the pitts model nand gate?
(C) Cognition
(A) 0
(D) Model Building
(B) 1
(C) either 0 or 1 17. You have an input volume
that is 32x32x16, and ap-
(D) z
ply max pooling with a
12. Which of the following is stride of 2 and a filter size
not a categories in learning of 2. What is the output vol-
of ANN ume?
(A) Enforcement learning (A) 15x15x16
(B) supervised learning, (B) 16x16x16
(C) unsupervised learning, (C) 32x32x8
(D) reinforcement learning (D) 16x16x8

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 37

18. Which of the following is a 23. Given a color image of size

learning rule in NN 120 x 80 pixels, what is the
(A) Hebeen learning rule. dimension of its feature vec-
tor?
(B) Perception learning rule.
(A) 9, 600
(C) Start Delta learning rule.
(B) 28, 800
(D) Competitive learning rule.
(C) 8, 400
19. s Problem Canvas (D) 25, 200
(A) make many lives better and 24. is an algorithm for pat-
help our country achieve these tern recognition by Rosen-
goals. blatt in 1958.
(B) helps in evaluating the solu- (A) Megatron
tion related to the problem
(B) Perceptron
(C) helps in identifying the key el-
ements related to the problem. (C) Vultron

(D) All of these (D) Terceptron

25. Backpropagation calculates

20. Each layer in NN is con-
the and propagates it
nected to the next layer
back to earlier layers.
through:
(A) value
(A) bias
(B) error
(B) neuron
(C) data
(C) weight
(D) none of above
(D) activation function
26. Please Rate this Presenta-
21. Domain of the AI used by tion.
price comparison websites
is (A) Best

(A) Data Science (B) Better than Ashley

(B) NLP (C) OK

(D) Needs work
(C) CV
(D) none of above 27. Function of dendrites is?
(A) receptors
22. To activate serial monitor,
what setup code is to be (B) transmitter
written? (C) both receptor & transmitter
(A) serialBegin(9600); (D) none of the mentioned
(B) Serialbegin(9600);
28. Which of the following is
(C) Serial.begin(9600); not a structure of a neuron?
(D) Serial.Begin(9600); (A) Dendrite

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
38 Chapter 2. Application of supervised learning

(B) Axon 33. Several sets of data related

(C) Newtron to each other used to make
decisions in machine learn-
(D) none of above ing algorithms. E.g. com-
ments to make you happy
29. The fundamental unit of net-

NARAYAN CHANGDER
and comments to make you
work is sad
(A) brain (A) Dataset
(B) neuron (B) unsupervised learning
(C) nucleus (C) supervised learning
(D) axon (D) Classifiers

30. WHY DO WE NEED BIOLOG- 34. What is ANN

ICAL NEURAL NETWORKS? (A) Artificial Neural Node
(A) TO SOLVE TASKS LIKE MA- (B) Artificial Neuro Nest
CHINE VISION ANDNATURAL (C) Artificial Neural Network
LANGUAGE PROCESSING
(D) Artificial Neuron Naive
(B) TO MAKE HUMAN INTERAC-
TIVE AND FRIENDLY 35. You have an input volume
that is 63x63x16, and con-
(C) TO APPLY HEURISTIC volve it with 32 filters that
SEARCH METHODS TO FIND SO- are each 7x7, using a stride
LUTIONS OF PROBLEM of 2 and no padding. What
(D) ALL OF THE ABOVE is the output volume?
(A) 16x16x32
31. The soft computing exploits
the tolerance of: (B) 29x29x16
(A) Imprecision (C) 29x29x32
(D) 16x16x16
(B) Uncertainty
(C) Partial truth 36. what is the ANN equivalent
to BNN synapses?
(D) Analytical model
(A) Node
32. which of the following is not (B) Input
a network architecture
(C) Weights
(A) Single-Layer Feed-Forward (D) Output
Network
37. What is shape of dendrites
(B) Multi-Layer Feed-Forward Net-
like
work
(A) oval
(C) Single Node Without Feed-
back network (B) round
(D) Multi-Layer Recurrent Net- (C) tree
work (D) rectangular

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 39

38. To predict a test data with 42. Which is the best output
a perceptron in python we configuration for a model
use tasked with predicting a pa-
(A) perceptron.fit(X train) tient’s age given their brain
MRI image?
(B) perceptron.fit(X train, y
(A) One neuron with sigmoid
train)
(B) Multiple neurons with softmax
(C) perceptron.predict(X test)
(D) perceptron.predict(X test, y (C) One neuron with linear output
test)

39. What is the trend in soft- (D) Multiple neurons with Tanh
ware nowadays?
43. A dense multi layer percep-
(A) to bring computer more & tron layer has 300 input
more closer to user values and 200 output val-
(B) to solve complex problems ues. Assuming no bias, how
many parameters does the
(C) to be task specific layer contain?
(D) to be versatile (A) 500
40. exploit spatially-local cor- (B) 6000
relation by enforcing a lo- (C) 60000
cal connectivity pattern be- (D) 3000
tween neurons of adjacent
layers. 44. What are dendrites?
(A) RNN or Recurrent Neural Net- (A) fibers of nerves
work (B) nuclear projections
(B) CNN or Convolution Neural (C) other name for nucleus
Network (D) none of the mentioned
(C) LSTMor Long-Short Term
Memory 45. What was the 2nd stage in
perceptron model called?
(D) none of above
(A) sensory units
41. Targetted marketing, Rec- (B) summing unit
ommended Systems, and (C) association unit
Customer Segmentation are
applications in (D) output unit

(A) Unsupervised Learning:Clustering 46. Which is the best output

configuration for a model
tasked with classifying a
(B) Supervised Learning:Classification
person’s mood, e.g. angry,
happy, sad etc. by the tone
(C) Reinforcement Learning of their voice?
(D) Unsupervised Learning:Regression (A) One neuron with sigmoid

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
40 Chapter 2. Application of supervised learning

(B) Multiple neurons with softmax 51. Every Artificial Neural must
have at least
(C) One neuron with linear output (A) two layers.
(B) three layers.

NARAYAN CHANGDER
(D) None of these (C) four layers.
47. short for Robotic Exoplanet (D) none of above
Recognition for Exoplanets
that are beyond our solar 52. It is the ability of a dig-
system. ital computer or computer-
controlled robot to perform
(A) Roberta tasks commonly associated
(B) Robert with intelligent beings
(C) Robot (A) Automated computations

(D) Robet (B) IOT

(C) Artificial Intelligence
48. What 2 subjects are neu-
ral networks usually associ- (D) None of these
ated with? 53. What does np.zero() do?
(A) Geology & CHemistry (A) Creates a matrix filled with ze-
(B) Geology & Biology roes
(C) Chemistry & Artificial- (B) Squares the input
engineering (C) Takes a list as input, and
(D) Biology & Artificial-engineering replaces negative numbers with
zero
49. WHAT PROBLEMS DO ANNs (D) Takes two lists as input, and re-
NOT SOLVE? turns the matrix product of them
(A) GEOMETRIC PROBLEMS both
(B) SCIENCE PROBLEMS 54. This is where we insert the
(C) INDUSTRY PROBLEM initial data for the neural
network.
(D) FINANCIAL PROBLEM
(A) Input
50. Which is true for neural net- (B) Hidden Layer
works?
(C) Output
(A) It has set of nodes and connec-
tions. (D) none of above

(B) Each node computes it’s 55. Which one of these is not an
weighted input. area of AI?
(C) Node could be in excited state (A) computer vision/image recog-
or non-excited state. nition
(D) All of the mentioned. (B) voice recognition

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 41

(C) web design (C) threshold

(D) robotics (D) synapse

56. What futuristic actions can 60. The helps us to sum-

not be performed by neural marise all the key points of
networks? Problem Scoping.
(A) Pattern Recognition (A) Problem Statement Template
(B) Function Approximation (B) 4 Ws Problem canvas
(C) Pattern Classification (C) Data Features
(D) Facial Recognition (D) none of above

57. Under this block, you also 61. they perform the same task
gather evidence to prove for every element of a se-
that the problem you have quence, with the output be-
selected actually exists. ing depended on the previ-
Identify the block of 4W ous computations.
problem canvas. (A) RNN or Recurrent Neural Net-
(A) Who work

(B) what (B) CNN or Convolution Neural

Network
(C) where
(C) LSTMor Long-Short Term
(D) why Memory
58. Assume a simple MLP (D) none of above
model with 3 neurons and
62. What does TPU stands for?
inputs= 1, 2, 3. The weights
to the input neurons are 4, 5 (A) Training Processor Unit
and 6 respectively. Assume (B) Training Processing Unit
the activation function is a
(C) Tensor Processor Unit
linear constant value of 3.
What will be the output? (D) Tensor Processing Unit
(A) 32 63. The most common Neural
(B) 96 networks consist of three
network layers:input layer,
(C) 48
output layer and
(D) none of above
(A) neuron layer
59. Location (space) where a (B) hidden layer
neuron can transfer an im- (C) algorithm layer
pulse to another cell with
the use of neurotransmitters (D) none of above
is called 64. Which is the correct struc-
(A) myelin sheath ture of a Neural Network?
(B) axon (A) Output, Hidden Layer, Input

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
42 Chapter 2. Application of supervised learning

(B) Hidden Layer, Input, Output 69. In learning only inputs

provided but no output re-
(C) Input, Hidden Layer, Output
lated data.
(D) none of above
(A) Supervised
65. Which of the following is (B) Reinforcement

NARAYAN CHANGDER
not the promise of artificial (C) Un supervised
neural network?
(D) Semi-Supervised
(A) It can explain result
70. What are ANNs used for?
(B) It can survive the failure of
some nodes (A) Reproduce Human Spinal
Functions
(C) It has inherent parallelism
(B) Reproduce Human Foot Func-
(D) It can handle noise tions
66. is a series of algorithms (C) Reproduce Human Nervous
that aims to understandthe System
relationships in a set of data (D) Reproduce Human Brain
using aprocess inspired by Functions
the human brain
71. What are the Advantages
(A) artificial neural network of Neural Networks?
(ANN)
(A) It can be performed without
(B) neuron (nerve cell) any problem
(C) K-nearest neighbor (B) It can be implemented in any
(D) k-mean value application.
(C) A neural network learns and
67. What futuristic actions reprogramming is not necessary
can not be performed by
ANNs? (D) All of the above

(A) Pattern Recognition 72. If a neural network has a

single output neuron, then
(B) Function Approximation
the model may be used for:
(C) Pattern Classification (A) Binary classification
(D) Facial Recognition (B) Regression
68. A set of instructions to fol- (C) Binary classification or regres-
low in order to solve a prob- sion
lem (D) None of these
(A) Algorithm 73. A 4-input neuron has
(B) Machine Learning weights 1, 2, 3 and 4. The
transfer function is linear
(C) Neuron
with the constant of propor-
(D) Noural Network tionality being equal to 2.

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 43

The inputs are 4, 10, 5 and 78. receive signal from other
20 respectively. What will neurons.
be the output? (A) Axon
(A) 238 (B) Dendrites
(B) 76 (C) nodes
(C) 119 (D) none of above
(D) 123
79. The total number of training
74. Neural Network is type of examples present in a single
which is type of machine batch
learning (A) Epochs
(A) natural language processing (B) Batch
(B) deep learning (C) Iteration
(C) speech (D) none of above
(D) robotics 80. What is an activation
value?
75. of data created is
(A) weighted sum of inputs
(A) Unstructured
(B) threshold value
(B) Minimally structured
(C) main input to neuron
(C) Structured
(D) none of the mentioned
(D) Randomized
81. Which of the following is
76. Neural Networks are com- the feature of a neural net-
plex with many parame- work?
ters.
(A) Extract information without
(A) Linear Functions any programming
(B) Nonlinear Functions (B) All of these
(C) Discrete Functions (C) Learn itself and produced the
(D) Exponential Functions trained data
(D) Provide the filtered informa-
77. What two types of Neural
tion without input
Networks are there
(A) Biological Neural network 82. What is the name of the test
and Artificial Neural Network that Alan created to check
whether something is hu-
(B) Chemical Neural Network man or a computer?
and Biological Neural Network
(A) Test of Turing
(C) Geological Neural Network
and Artificial Neural Network (B) Turner Test

(D) Chemical Neural Network (C) Turing Machine

and Geological Neural Network (D) Turing Test

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
44 Chapter 2. Application of supervised learning

83. Who invented perceptron 88. WHAT ARE ANNs USED

neural networks? FOR?
(A) McCullocch-pitts (A) PERFORM HUMAN NERVOUS
(B) prince SYSTEM

NARAYAN CHANGDER
(C) Minsky & papert (B) PERFORM HUMAN SPINAL
FUNCTIONS
(D) Rosenblatt
(C) PERFORM HUMAN BRAIN
84. which of the following is not FUNCTIONS
an activation function of NN
(D) PERFORM HUMAN EYE
(A) Sigmoidal Function FUNCTION
(B) Hyperbolic Tangent Function
(Tanh) 89. Which layer performs com-
putation in a multilayer per-
(C) Triangular Function
ceptron?
(D) Rectified Linear Unit (ReLU)
(A) input nodes
Function
(B) Auxiliary nodes.
85. You have an input volume
that is 15x15x8, and pad it (C) hidden nodes
using “pad=2.” What is the (D) none of above
dimension of the resulting
volume (after padding)? 90. The actual processing oc-
curs in which of the follow-
(A) 19x19x12
ing layer?
(B) 17x17x10
(A) input layers
(C) 19x19x8
(B) hidden layers
(D) 17x17x8
(C) output layers
86. computational model in- (D) None of these
spired by the structure and
functions of the biological 91. What Problems do ANNs
neural network not solve?
(A) artificial neural network (A) Geometric problems
(ANN)
(B) Science problems
(B) neuron
(C) Industry problems
(C) weight
(D) Financial problems
(D) bias
92. It is a subset of Artifi-
87. Neural network is inspired
cial Intelligence which en-
by
ables machines to improve
(A) Human heart at tasks with experience
(B) Human lungs (data).
(C) Human brain (A) Machine Learning
(D) human eyes (B) AI

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 45

(C) Deep Learning 97. Identify the correct order of

the 4Ws problem canvas.
(D) none of above
(A) Who? Why? When? Where?
93. The part of the neuron that (B) Who? What? Where? Why?
brings information to the
cell body is called a(n) (C) Why? What? Where Who?
(D) Why? What? Who? Where?
(A) dendrite
(B) axon 98. This is a system of Pro-
grams and Data Structures
(C) nucleus that mimics the operation of
(D) node of Ranvier the human brain:
(A) Intelligent Network
94. The network that involves
backward links from output (B) Decision Support System
to the input and hidden lay- (C) Neural Network
ers is called (D) Genetic Programming
(A) Self organizing maps
99. What was Alan Turing’s pro-
(B) Perceptrons fession before he worked
(C) Recurrent neural network for the government?

(D) Multi layered perceptron (A) Mechanic

(B) Mathematician
95. Artificial pattern-recognition
(C) Nurse
algorithms achieve human-
level performance on cer- (D) Teacher
tain tasks by this year.
100. The activation of each neu-
(A) 1202 ron in a layer is determined
by the of the activations
(B) 2002
of all the neurons in the pre-
(C) 2001 vious layer.
(D) 2012 (A) weighted sum
(B) sum
96. In this type of model, the al-
gorithms work towards ac- (C) level
complishing the goal or try (D) none of above
to improve the performance
in a particular task. This is 101. Suppose your input is a
used in gaming. 300 by 300 color (RGB) im-
age, and you use a convo-
(A) Reinforcement learning lutional layer with 100 fil-
(B) Supervised Learning ters that are each 5x5. How
many parameters does this
(C) UnSupervised Learning
hidden layer have (without
(D) none of above bias)

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
46 Chapter 2. Application of supervised learning

(A) 2501 106. What is NOT the property

(B) 2600 of sigmoid function?

(C) 7500 (A) Input is any real number

(D) 7600 (B) Output is a real number in

range between 0 to 1

NARAYAN CHANGDER
102. It is an application of ar- (C) Nearly linear when input is
tificial intelligence (AI) that close to zero
provides systems the abil-
ity to automatically learn (D) Slope is close to zero when in-
and improve from experi- put is close to zero
ence without being explic- 107. A 4-input neuron has
itly programmed weights 1, 2, 3 and 4. The
(A) Artificial Intelligence activation function is linear
with the constant of pro-
(B) Decision Tree
portionality being equal to
(C) Machine Learning 2 (y=2x). The inputs, Ox
(D) Linear Regression are 4, 10, 5 and 20 respec-
tively. What will be the out-
103. The current Artificial Neu- put, Oy?
ral Networks are part of
(A) 238
(A) Weak Al systems.
(B) 76
(B) Strong Al systems.
(C) 119
(C) Future Al systems.
(D) 123
(D) none of above
108. What would be consid-
104. collection of ‘neurons’ op- ered stimuli?
erating together at a specific
(A) Friends in the classroom
depth within a neural net-
work (B) Informational posters on the
wall
(A) Bias
(C) Models that are hands on
(B) Weight
(D) All of the above
(C) Neuron
(D) Layers 109. What’s the main point of
difference between human
105. What if the learning rate is & machine intelligence?
too high? (A) human perceive everything as
(A) Model is difficult to converge a pattern while machine perceive
(B) Model consumes more com- it merely as data
putational resources (B) human have emotions
(C) Model takes longer time to (C) human have more IQ & intel-
converge lect
(D) Model is difficult to generalize (D) human have sense organs

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 47

110. What is delta (error) in per- (A) Algorithm + business process

ceptron model of neuron?
(A) error due to environmental (B) Data + business process
condition (C) Algorithm + data
(B) difference between desired & (D) Algorithm + data science
target output
(C) can be both due to difference 115. What is the capacity of a
in target output or environmental neural network?
condition (A) number of outputs
(D) none of the mentioned (B) number of inputs
111. In this model, the ma- (C) number of patterns it can
chine is trained with huge store
amounts of data which (D) none of above
helps it in training itself
around the data. Such 116. The AI domain which can
machines are intelligent be used to predict AIR qual-
enough to develop algo- ity index is
rithms for themselves. (A) NLP
(A) AI (B) CV
(B) ML (C) Data Science
(C) DL (D) none of above
(D) none of above
117. The number of nodes in in-
112. WHO PROPOSED FIRST put layer is 6 and the Hid-
COMPUTATIONAL MODEL den layer is 10 the maxi-
OF A NEURON? mum number of connections
from the input layer to the
(A) ROSENBLATT
hidden layer is:
(B) McCULLOCH-PITTS
(A) 60
(C) MINSKY & PAPERT
(B) 30
(D) MAcCARTHY
(C) 90
113. SUPERVISED LEARNING IN- (D) none of above
CLUDES WHICH TASK?
118. The number of batches
(A) CLASSIFICATION
needed to complete one
(B) REGRESSION epoch
(C) BOTH 1 & 2 (A) Iterations
(D) NONE OF THESE (B) Epochs

114. Machine Learning model is (C) Batch

(D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
48 Chapter 2. Application of supervised learning

119. Domain of the game Rock (D) All of them

Paper Scissors is
(A) NLP 124. Mean squared error is a
common loss function for
(B) Data Science which task?

NARAYAN CHANGDER
(C) Computer Vision
(A) Regression
(D) none of above
(B) Classification
120. If we want to create a
model of neural network (C) None of these
that can identify whether (D) Regression and Classification
you are wearing a mask or
no, what is the ideal number 125. What are the minimum
of output that we should cre- number of layers we can
ate? have in a NN
(A) 1
(A) Four
(B) 2
(B) Five
(C) 3
(D) none of above (C) Three
(D) Two
121. This is an intermediate
layer between input and
126. In which of the following
output layer and place
type of learning networks
where all the computation
are trained to provide the
is done.
correct output using several
(A) Input example units?
(B) Hidden Layer
(A) Supervised learning
(C) Output
(B) Unsupervised learning
(D) none of above
(C) Reinforcement learning
122. It produces the result for
given inputs. (D) All of these
(A) Input
127. Why do we need biologi-
(B) Hidden Layer cal neural networks?
(C) Output (A) to solve tasks like machine vi-
(D) none of above sion & natural language process-
ing
123. What is the correct exam-
ple of Supervised learning? (B) to apply heuristic search meth-
(A) Face recognition ods to find solutions of problem

(B) Speech Recognition (C) to make smart human interac-

tive & user friendly system
(C) Netflix and YouTube recom-
mendations (D) all of the mentioned

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.8 NEURAL NETWORK 49

128. What if we would like (C) One iteration

to have prediction output
(D) none of above
(binary classification) rep-
resented by probability, 133. The number of nodes in
which activation function is the input layer is 10 and the
the best choice? hidden layer is 5. The maxi-
(A) tanh mum number of connections
(B) ReLu from the input layer to the
hidden layer are
(C) Sigmoid
(A) 50
(D) Linear
(B) 40
129. what is the BNN equiva-
(C) 60
lent to ANN output?
(A) Soma (D) none of above

(B) Dendrite 134. *In this type of model,

(C) Synapses the algorithm learns and
makes sense by extracting
(D) Axon
features/patterns from the
130. WHAT IS THE FUNCTION unlabelled dataset provided
OF AN AXON? (The system will evaluate by
itself)
(A) TRANSMISSION
(A) Unsupervised learning
(B) TRANSMITTER
(B) supervised learning
(C) RECEPTION
(D) NONE OF THESE (C) Reinforcement learning
(D) none of above
131. What does lower latency
mean? 135. Example of a unsuper-
(A) Less time spent looking for net- vised feature map?
works (A) text recognition
(B) Faster Facebook scrolling
(B) voice recognition
(C) Less time for information to
travel to and from your device (C) image recognition

(D) Enhanced computing perfor- (D) none of the mentioned

mances
136. Exmples of Machine learn-
132. When an ENTIRE dataset is ing
passed forward and back- (A) All of Them
ward through the neural net-
work only ONCE (B) Speech to text
(A) One epoch (C) Google maps
(B) One batch (D) Siri, Uber

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 50 Chapter 2. Application of supervised learning

137. To train a perceptron in train)

python we use (C) perceptron.predict(X test)
(A) perceptron.fit(X train) (D) perceptron.predict(X test, y
(B) perceptron.fit(X train, y test)

NARAYAN CHANGDER
2.9 Deep Learning
1. Given a sound to clip of a (C) Long Short Term Memory Net-
person or people speaking, works
determining the textual rep-
(D) Gated Recurrent Unit Net-
resentation of the speech is
works
(A) Speech Synthesis
(B) Speech Recognition 5. Which of the following tech-
nique is also called as learn-
(C) Prediction of a speech ing from teacher?
(D) All of the above (A) Unsupervised Learning
2. If the number of classes (B) Supervised Learning
present in the data is two,
(C) Reinforcement Learning
which of the following can
be used to measure loss (D) All of the Above
(A) Binary cross entropy 6. Artificial Intelligence is su-
(B) Categorical cross entropy perset of & ,
(C) Root mean squared error (A) Machine Learning & Neural
(D) Sum of squared error Networks
(B) Machine Learning & Deep
3. Input layer contains “n” in-
Learning
puts, hidden layer has “h”
number of nodes. What (C) Deep Learning & Neural Net-
is the dimensionality of works
weight matrix (consider (D) none of above
presence of “bias term”)?
(A) (n+1) x h 7. What are general limita-
tions of back propagation
(B) n x (h+1)
rule? Pick the right choice
(C) n x (h-1) from the given options i) lo-
(D) (n-1) x h cal minima problem ii) slow
convergence iii) scaling
4. occurs when the gradi- (A) only i
ents become very small and
tend towards zero. (B) both ii and i
(A) Exploding Gradients (C) both ii and iii
(B) Vanishing Gradients (D) i, ii and iii

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.9 Deep Learning 51

8. which CNN has more lay- (C) Neural Networks

ers?
(D) none of above
(A) Resnet-101
13. Model Ensembling is:
(B) Lenet
(C) Googlenet (A) Having multiple instances of
the network(s) and average to-
(D) Alexnet gether their responses
9. Which of the following is cor- (B) Having a single instance of the
rect formula for calculating network and pass the input mul-
convolution output? tiple times but altered in a small
way
(A) (W-2P + F)/S + 1
(C) The perfect string quartet
(B) (W + 2F-S)/P + 1
(C) (W + 2P-F)/S + 1 (D) None of the above

(D) (W + 2S-F)/P + 1 14. Each cell in LSTM Network

Consists of number of
10. Which Parameter need to Gates
be learnt in minimizing ob-
jective function in super- (A) 2
vised learning. (B) 3
(A) Only weight (C) 4
(B) Only Bias (D) 5
(C) Both weight and bias
15. What is function which re-
(D) Learning rate
flects the error between the
11. If training accuracy is well target output and actual out-
above validation accuracy put of the perceptron?
in all epochs, it is an indica- (A) Perceptron function
tion of model
(B) Actual function
(A) Underfitting
(C) Output function
(B) Overfitting
(D) Loss function
(C) Correct fit
(D) None of the above 16. Correct tensor flow function
to restore model
12. This requires computer sci-
(A) tf.keras.models.load
entists to formulate general-
model
purpose learning algo-
rithms that help machines (B) tf.keras.models.restore
learn more than just one model
task. (C) tf.keras.models.save
(A) Machine Learning (D) tf.keras.models.save re-
(B) Deep Learning store

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
52 Chapter 2. Application of supervised learning

17. Backpropagation works by (C) Ideas or objects

first calculating the gradient (D) All of the choices
of and then propagating
it backwards. 22. uses the processing of
(A) Sum of squared error with re- the brain as a basis to de-

NARAYAN CHANGDER
spect to inputs velop algorithms that can be
used to model complex pat-
(B) Sum of squared error with re- terns and prediction prob-
spect to weights lems.
(C) Sum of squared error with re- (A) ANN
spect to outputs
(B) CNN
(D) None of the above
(C) RNN
18. Which of the following is (D) KNN
a representation learning al-
gorithm? 23. A machine learning system
(A) Neural network is rather than
(B) Random Forest (A) untrained, implicitly pro-
grammed
(C) k-Nearest neighbor
(B) Publicly trained, privately pro-
(D) None of the above
grammed
19. Which of the following deep (C) Trained, explicitly pro-
learning models uses back grammed
propagation?
(D) none of above
(A) Convolutional Neural Net-
work 24. Why is the batch descent
algorithm not commonly
(B) Multilayer Perceptron Network
used?
(A) Too many hidden layers
(C) Recurrent Neural Network
(B) Hard to be implemented
(D) All of the above
(C) Too complex
20. How many layers Deep
(D) The convergence process is
learning algorithms are con-
very slow
structed?
(A) 2 25. What is the objective
of backpropagation algo-
(B) 3
rithm?
(C) 4
(A) to develop learning algorithm
(D) 5 for multilayer feedforward neural
network
21. What is data?
(B) to develop learning algorithm
(A) Unprocessed infomation for single layer feedforward neu-
(B) Set of values ral network

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.9 Deep Learning 53

(C) to propagate the computed (B) Cannot handle very large data
loss to compute gradient
(D) all the mentioned (C) Assume normal distribution
26. Which of the following loss (D) Assume independent of fea-
functions is best for Classifi- tures
cation?
31. A decision boundary in an
(A) L1
unspecified number of di-
(B) L2 mensions is called
(C) Manhattan Distance (A) Line
(D) Negative Loglikelihood (B) Plane
27. When the dataset is small (C) Hyperline
which of the following will
(D) Hyperplane
give better performance?
(A) Deep learning 32. Which is the following is
(B) Machine learning true about neurons?

(C) Both a and b (A) A neuron has a single input

and only single output
(D) None of the above
(B) A neuron has multiple inputs
28. Which of the following neu- and multiple outputs
ral network training chal-
lenge can be solved using (C) A neuron has a single input
batch normalization? and multiple outputs

(A) Overfitting (D) All of the above

(B) Restrict activations to become 33. What are machine learning
too high or low Models all about
(C) Training is too slow
(A) transformation of data
(D) All the mentioned
(B) representations of data
29. Which is the best descrip- (C) Appropriate representations
tion of Prescriptive Analyt- for there Input data
ics?
(D) none of above
(A) What to do
(B) Why to do 34. Sentiment analysis is an ex-
ample of this process se-
(C) Predict the possibility
quence
(D) Tell what to do
(A) One to one
30. What is the limitation of per-
(B) One to many
ceptron?
(C) Many to one
(A) Cannot solve non-linearly sep-
arable problem (D) Many to many

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
54 Chapter 2. Application of supervised learning

35. The process of evaluating 40. Number of output cases de-

data through analytical and pends on what factor?
statistical tools (A) number of inputs
(A) Data Mining (B) number of distinct classes
(B) Data Analysis

NARAYAN CHANGDER
(C) total number of classes
(C) Data Exploration (D) none of the mentioned
(D) Data Visualization
41. Which of the following is
36. Which of the following is a disadvantage of decision
a low-level programming trees?
library for implementing (A) Factor analysis
deep neural networks?
(B) Decision trees are robust to
(A) Tensorflow outliers
(B) Python (C) Decision trees are prone to be
(C) Keras overfit
(D) R (D) None of the above

37. A AI pioneer said that 42. Which function is often used

perceptron cannot process as the output layer of a mul-
non-linear data. Who is ticlass classification task?
he/she? (A) Softmax
(A) Turing (B) Softplus
(B) Simon (C) Softsign
(C) Shannon (D) Softmin
(D) Minsky
43. Deep learning is successful
38. Which of the following tech- due to
nique does not belongs to (A) Power to estimate huge num-
Deep learning ber of parameters
(A) Convolution Neural Network (B) Power to express linear and
(B) Support Vector Machine non-linear relationship.
(C) Transfer Learning (C) Availability of high computa-
(D) Gated Recurrent Unit tional resources.
(D) All of these
39. Which of the following is
a subset of machine learn- 44. What is the advantage of
ing? deep learning?
(A) Pandas (A) Poor performance
(B) Pycharm (B) Improved Error
(C) Deep learning (C) Best Performance
(D) none of above (D) None of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.9 Deep Learning 55

45. Which of these are categor- 50. What is the definition of AI

ical features? (A) The effort to automate intellec-
(A) Height of a person tual tasks normally performed by
humans
(B) Price of petroleum
(B) The effect to understand intel-
(C) Mother tongue of a person
lectual tasks normally performed
(D) Amount of rainfall in a day by humans

46. Which of the following (C) The effort to automate basic

gives non-linearity to a neu- day to day tasks normally per-
ral network? formed by humans

(A) Stochastic Gradient Descent (D) none of above

(B) Rectified Linear Unit 51. Which of the following algo-

rithm is not an example of
(C) Convolution function
an ensemble method?
(D) None of the above
(A) Random Forest
47. Which of the following is (B) Gradient Boosting
not an activation function?
(C) Decision Tree
(A) ReLU
(D) None of these
(B) Tanh
52. Which of the following can
(C) Sigmoid be used for machine transla-
(D) SGD tion?
(A) CNN
48. Which is true about ensem-
bles? (B) MLP
(A) Reduce variance error (C) Perceptron
(B) Reduce bias error (D) RNN
(C) Surely, reduce variance error 53. Which of the following
(D) none of these layer is not part of CNN Ar-
chitecture?
49. What does the deep in deep (A) POOL
learning stand for?
(B) CONV
(A) it stands for this idea of regres-
sivelayers of representations. (C) FC

(B) it stands for this idea of (D) SOFTMAX

progressives’layers of representa-
54. Correct tensor flow class to
tions. How
create fully connected layer
(C) it stands for this idea of suc- in a deep neural network
cessivelayers of representations.
(A) tensorflow.keras.layers.Flatten
(D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
56 Chapter 2. Application of supervised learning

(B) tensorflow.keras.layers.Dense 59. The Process of adjusting

weight is known as
(C) tensorflow.keras.Sequential (A) Activation
(D) tensorflow.keras.Model (B) Synchronisation

NARAYAN CHANGDER
55. Which of the following (C) Learning
more number of computa- (D) none of the mentioned
tion cores.
60. Which of the folllowing is
(A) CPU
an example of feature ex-
(B) GPU traction?
(C) AMD (A) Constructing bag of words
(D) ARM vector from an email

56. Which of the following func- (B) Applying PCA projects to a

tions can numerically sta- large high-dimensional data
bilize overflow and under- (C) Removing stopwords in a sen-
flow? tence
(A) Softminux function (D) All of the above
(B) Softplus function
61. What is class conditional in-
(C) Softmax function dependence?
(D) Softmin function (A) Classes in the data are inde-
pendent of one another
57. Which of the following can
act as possible termination (B) Classes in the data are depen-
conditions in K-Means? dent on one another
(A) For a fixed number of itera- (C) Features in the data are inde-
tions. pendent of one another
(B) Assignment of observations to (D) Features in the data are inde-
clusters does not change between pendent of one another given the
iterations. class
(C) Terminate when residual sum 62. is used to find local min-
of squares falls below a threshold ima of the cost function
(A) stochastic gradient descent
(D) all the above
(B) gradient descent
58. Classifying an email into
(C) linear regression
spam, work related and pri-
vate comes under? (D) logistic regression
(A) Descriptive task 63. Which of the following lay-
(B) Binary classification model ers is NOT a part of CNN?
(C) Regression model (A) Convolutional Layer
(D) Multiclass classification model (B) Pooling Layer

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.9 Deep Learning 57

(C) Code Layer (C) Mini batch gradient descent

(D) Fully connected Layer (D) All of the above

64. What is the filter Size used 69. Feedback networks are
in AlexNet Architecture? used for?
(A) 3 (A) auto association
(B) 5 (B) pattern storage
(C) 7 (C) both auto association & pat-
tern storage
(D) 11
(D) pattern recognition
65. Which of the following is a
type of neural network? 70. Which of the following is an
example of deep learning?
(A) RNN (Recurrent Neural Net-
work) (A) Self-Driving cars
(B) LSTM (Long Short-Term Mem- (B) Pattern Recognition
ory Networks) (C) Natural Language Processing
(C) Restricted Boltzmann Ma-
chines (RBM) (D) All of the above
(D) All of the above 71. Batch normalization helps
66. In VC dimensions of neu- to prevent
ral networks what does VC (A) activation functions to become
stand for? too high or low
(A) Vladimir Cherubim (B) the training speed to become
(B) Vapnik Chervonenkis too slow

(C) Victor Charlie (C) Both A and B

(D) Vanessa Carlton (D) None

72. Which of the following is an
67. A = 1/(1 + e-x) is an equa-
optimizer?
tion representing which acti-
vation function? (A) Categorical cross entropy
(A) ReLU (B) Accuracy
(B) Sigmoid (C) Binary cross entropy
(C) Leaky ReLu (D) Adam
(D) Tanh 73. In CNN’s which of the fol-
lowing layers exploits spa-
68. If we want our neural net-
tial locality of input data?
work to converge faster
which of the following (A) Fully connected layer
should be used (B) Input layer
(A) Batch gradient descent (C) Convolution layer
(B) Stochastic gradient descent (D) Max pooling layer

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
58 Chapter 2. Application of supervised learning

74. Which of the following is an 79. The neural net models use
application of RNN? and principles to mimic
(A) Natural Language Processing the processes of the human
brain, allowing for more
(B) Audio and Video analysis general learning.

NARAYAN CHANGDER
(C) Stock market prediction
(A) Maths & Science
(D) All of the above
(B) Maths & Computer Science
75. The price of a smartphone
(A) Cluster (C) Human Brain & Science
(B) Class (D) none of above
(C) Regression
(D) none of above 80. CNN is best suited for:

76. Machine Learning is a sub- (A) Image Classification

set of AI. ML deals with de- (B) Natural Language Processing
veloping systems which can
improve their performance (C) Image Captioning
with
(D) All of the above
(A) Experience
(B) Deep Learning 81. Which of the following does
(C) Neural Networks not come under text analy-
sis?
(D) none of above
(A) Document Structure Detection
77. Which of the following is an
application of reinforcement
learning? (B) Speech Synthesis
(A) Predicting the stock price in
(C) Prosonic Analysis
the next month
(B) Self-driving cars (D) Prosodic Analysis
(C) Predicting whether a customer
will buy new model phone 82. For an image recognition
problem (recognizing a cat
(D) Recommending new items in a photo), which archi-
based on past purchase history tecture of neural network
of the customer would be better suited to
78. How many learnable pa- solve the problem?
rameters does a pooling (A) Multi Layer Perceptron
layer have?
(A) 1 (B) Convolutional Neural Net-
work
(B) 0
(C) 2 (C) Recurrent Neural network
(D) 3 (D) Perceptron

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.9 Deep Learning 59

83. To find the minimum or the (C) Vanishing gradient

maximum of a function, we
(D) Dropout
set the gradient to zero be-
cause: 88. The measure of the propor-
(A) The value of the gradient at ex- tion of actual positive values
trema of a function is always zero that got predicted as posi-
tive is
(B) Depends on the type of prob- (A) Specificity
lem
(B) Sensitivity
(C) Both A and B
(C) Accuracy
(D) None of the above
(D) None of the above
84. Why is second order dif-
ferencing in time series 89. Which of the following func-
needed? tions can be used as an acti-
vation function in the output
(A) To remove stationarity
layer if we wish to predict
(B) To find the maxima or minima the probabilities of n classes
at the local point (p1, p2..pk) such that sum of
(C) Both A and B p over all n equals to 1?
(D) None of the above (A) Softmax

85. A Machine Learning method (B) ReLu

that is concerned with how (C) Sigmoid
software agents should
(D) Tanh
take actions in an environ-
ment is 90. What are the 3 main con-
(A) Reinforcement learning cepts of Data Science?
(B) semi-supervised learning (A) Data, Science, and Knowl-
(C) unsupervised learning edge
(D) supervised learning (B) Mathematics, Computer Sci-
ence, and Domain Expertise
86. CNN is mostly used when
there is an? (C) Machine Learning, Data Pro-
cessing, and Statistical Research
(A) structured data
(D) none of above
(B) unstructured data
(C) both the above 91. What is Big Data?
(D) None of the above (A) Data with a large size
87. These are the solutions for (B) Data with the word ‘big’ in it
overfitting (C) Data about people who are
(A) Softmax big
(B) Max pooling (D) Data made with a big purpose

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
60 Chapter 2. Application of supervised learning

92. What is true about an en- 97. The algorithm creates a line
sembled classifier? or a hyperplane which sep-
(A) Classifiers that are more arates the data into classes
“sure” can vote with more convic- and also suitable for classifi-
tion cation and regression

NARAYAN CHANGDER
(B) Classifiers can be more “sure” (A) K-means clustering
about a particular part of the (B) Support vector machine
space
(C) Bayesian inference
(C) Most of the times, it performs
better than a single classifier (D) perceptron
(D) All of the above
98. Why RNN doesn’t suit long-
93. Which of the following is term sequence data?
many to one prediction task
(A) Because it doesn’t have
in deep learning?
enough input
(A) Image caption generation
(B) None of DL models suit long-
(B) Text summarization term sequence data
(C) Sentiment analysis
(C) Due to vanishing gradient
(D) Text generation problem
94. Which of the following is a (D) SueSu said so
deep learning library?
(A) Tensorflow 99. LSTM is a variation of
(B) Keras (A) Convolutional Neural Net-
(C) PyTorch work
(D) All of the above (B) Recurrent Neural Network

95. How many max inputs can (C) Multi Layer Perceptron Net-
be given to one neuron? work
(A) 2 (D) None
(B) 5
100. Which of the following op-
(C) 36 tions is not the reason why
(D) None of the above traditional machine learn-
ing algorithms promote the
96. is set of algorithms that development of deep learn-
allows computers to learn ing?
from data without being ex-
plicitly programmed. (A) dimension disaster
(A) Machine Learning (B) local invariance and smooth
(B) Deep Learning regularization

(C) Neural Networks (C) manifold learning

(D) none of above (D) feature engineering

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.10 Perceptron 61

101. Generally, an ensemble 102. Cost function(J) of Linear

method works better, if Regression is the value
the individual base models between predicted y value
have accuracy (predicted) and true y value
(y)
(A) more than 60%
(A) Mean
(B) less than 60%
(B) Root mean square
(C) more than 50% (C) Median
(D) less than 50% (D) Mean square

2.10 Perceptron
2.11 Machine Learning Pipeline
1. A retail company wants to (A) Batch size
deploy a machine learning (B) Learning rate
model to predict the de-
mand for a product using (C) Initial weights
sales data from the past (D) Bias term
5 years.What is the MOST
efficient solution that the 4. You are preprocessing a
company should implement credit card fraud dataset
first? that contains the column
‘Amount’. You want to
(A) Regression
determine what the aver-
(B) Multi-class classification age amount is for the cases
(C) Binary class classification where there is fraud. Which
basic descriptive statistic
(D) none of above
could you use to achieve
2. Which feature of Amazon this?
SageMaker can you use to (A) Mode
learn patterns in data?
(B) Median
(A) SageMaker notebook in-
stances (C) Standard Deviation
(B) SageMaker training jobs (D) Mean
(C) SageMaker hyperaparameter 5. In which phase of the ML
tuning pipeline do you analyze the
(D) SageMaker endpoints business requirements and
re-frame that information
3. When updating your into a machine learning con-
weights using the loss func- text.
tion, what dictates how
much change the weights (A) Problem formulation
should have? (B) Model training

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 62 Chapter 2. Application of supervised learning

(C) Deployment in Amazon SageMaker to

(D) Data preprocessing solve this problem?
(A) Grid Search
6. When preparing the dataset
(B) Random Search
for your machine learning

NARAYAN CHANGDER
model, you should use one- (C) Bayesian optimization
hot encoding on what type (D) none of above
of data
8. Which of the following is
(A) Continuous
NOT an example of machine
(B) Ordinal learning?
(C) Nominal (A) Estimating home prices based
(D) Numerical on attributes of the property
(B) Sorting department store
7. A Data Scientist wants items by bar code
to tune the hyperparam-
eters of a machine learn- (C) Identifying objects in an im-
ing model to improve the age based on training data
model’s F1 score.Which tech- (D) Recommending products
nique CANNOT be used based on users’ buying habits

2.12 Artificial Intelligence

1. Which of the applications be- (B) Clustering and Dimensionality
low is prominent in AI? Reduction
(A) Speech Recognition (C) Rule Based and Learning
Based
(B) Honesty
(D) Classification and Clustering
(C) Emotional Management
(D) Anxiety 4. Which of these is NOT an AI
technology
2. Can be defined as the abil- (A) Facial recognition
ity of a computer to perform
tasks that require human in- (B) Image recognition
telligence. (C) Robotics
(A) Machine Learning (D) Animation
(B) Artificial Intelligence
5. What would be the output?
(C) Unsupervised Learning print(3 + 7)
(D) none of above (A) 10

3. Give 2 examples of Super- (B) 37

vised Learning models. (C) 3+7
(A) Classification and Regression (D) (3+7)

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 63

6. is a domain of AI re- (A) Yes

lated to data systems and (B) No
processes, in which the sys-
tem collects numerous data, (C) Kind of
maintains data sets and de- (D) none of above
rives meaning/sense out of
them. 11. Jobs that are highly repet-
itive and require weak
(A) NLP skills are the jobs most eas-
(B) ML-Data ily replaced by AI:
(C) Computer Vision (A) Computing
(D) none of above (B) Intelligent
(C) Social
7. When was AI first in-
vented? (D) Vocal
(A) 1920’s 12. AGI means
(B) 1950’s (A) Artificial intelligence
(C) 2000’s (B) Graphical artificial intelli-
(D) 1990’s gence
(C) Artificial Intelligence
8. Last month, we had our
representatives a confer- (D) Super intelligence
ence on environmental pol-
13. Ten million text messages
lution.
are sent on every
(A) to attend minute.
(B) attend (A) account
(C) attending (B) average
(D) attended (C) common
9. In artificial Intelligence, (D) general
knowledge can be repre-
14. The ability of Al that can be
sented as
used in voice operated re-
(A) Predicate Logic sponse systems is
(B) Propositional Logic (A) NLP
(C) Machine Logic (B) Text processing
(D) Both I and II (C) Speech recognition

10. Big data storage solutions (D) Visualisation

that can store huge amounts 15. What decade did formal A.I
of data (such as images or research begin?
video) and stream them to
many users at the same (A) 1980
time. Is this AI (B) 1950

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
64 Chapter 2. Application of supervised learning

(C) 1960 (C) Simple reflex AI agent

(D) 2000 (D) Unity-based AI agent

16. If people lose their jobs be- 21. Steps involved in the self
cause of AI, what will they driving cars:

NARAYAN CHANGDER
do? (A) Object detection
(A) become homeless (B) Motion planning
(B) retire from working (C) Both of the above
(C) stay home (D) None of the above
(D) learn how to do something dif- 22. CLOSEST in meaningRay
ferent Kurzweil suggests that ex-
ponential technological de-
17. This application technology velopment will lead to the
are expected to be the most inevitable rise of artificial in-
influential AI application in telligence.
future.
(A) inapplicable
(A) 5G Network (B) unavoidable
(B) Autonomous Vehicle (C) unrecognizable
(C) Computer Vision (D) invisible
(D) Block Chain 23. can be defined as the
18. AI is capable to process ability of a computer to per-
amount of data at higher form tasks that require hu-
speed. man intelligence.

(A) Tiny (A) Machine Learning

(B) Artificial Intelligence
(B) Enormous
(C) Unsupervised Learning
(C) Moderate
(D) Human intelligence
(D) All of these
24. LISP was created by?
19. I’m thinking of getting a fax
(A) John McCarthy
(B) Marvin Minsky
(A) engine
(C) Alan Turing
(B) instrument
(D) Allen Newell and Herbert Si-
(C) machine mon
(D) motor 25. Which one of these does not
20. Which of the following is use a form of AI?
not a type of Artificial Intelli- (A) Apple’s Siri
gence agent? (B) Amazon’s Alexa
(A) Learning AI agent (C) Android’s Google
(B) Goal-based AI agent (D) Microsoft Word

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 65

26. Investment and interest (C) NLP

in Al boomed in the first (D) Data Sciences
decades of the 21st cen-
tury, machine learning 31. What is the heuristic func-
was successfully applied to tion of greedy best-first
many problems in academia search?
and industry. (A) f(n) > h(n)
(A) while (B) f(n) != h(n)
(B) where (C) f(n) < h(n)
(C) which (D) f(n) = h(n)
(D) when
32. How many types of artifi-
27. What are the two main fea- cial intelligences are there?
tures of Big Data? (A) 8
(A) Volume (B) 5
(B) Speed (C) 3
(C) Both a) and b) (D) 1
(D) Accuracy
33. I’ll have a new dress for
28. Which of the these machine my daughter.
learning types uses a combi- (A) making
nation of labeled and unla-
beled data? (B) to make

(A) Supervised learning (C) make

(B) Semi-supervised learning (D) made

(C) Unsupervised learning 34. Many people question

(D) Reinforcement learning whether an AI can make eth-
ical decisions. What do we
29. Machine with intelli- mean by ETHICS?
gence works in a limited (A) The laws of particular country
functional area.
(A) Narrow (B) The rules of behaviour im-
(B) Weak posed by a programmer or the AI
(C) Both a) and b) creator

(D) None of these (C) A set of moral principles

(D) Human emotions
30. is the sub-field of AI that
is focused on enabling com- 35. I can solve complex prob-
puters to understand and lems that the human cannot
process human languages. solve
(A) Deep Learning (A) Narrow AI
(B) Machine Learning (B) General AI

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
66 Chapter 2. Application of supervised learning

(C) Super AI more suitable to solve this

(D) none of above problem?
(A) Perceptron
36. In programming, what is
the definition of precision? (B) RNN

NARAYAN CHANGDER
(C) CNN
(A) How quickly your program
can perform tasks (D) Fully connected neural net-
work
(B) How much you should trust the
accuracy of your program 41. If a computer can play chess
(C) How many math problems it known as
your program can complete in a (A) a good
given time
(B) a simple
(D) How well your program can
(C) an intelligent
perform on a video game
(D) none of above
37. Which of the given lan-
guage is not commonly used 42. Which of the following
for AI? statements are FALSE?
(A) LISP (A) An AI might be placed in a po-
sition where it has to make a life
(B) PROLOG or death decision
(C) Python (B) AI’s can learn from the mis-
(D) Perl takes that they make

38. What means was devised (C) AI’s use the same criteria as
to help work out whether humans, when making decisions
or not a computer can think (D) none of above
like a human?
43. refers to the AI modelling
(A) IQ test where the machine learns
(B) AI test by itself.
(C) Turing Test (A) Machine Learning
(D) Machine Learning (B) Learning Based
(C) Rule Based
39. Color recognition:
(D) Data Sciences
(A) computer vision
(B) robotics 44. OPPOSITE in meaningPro-
fessor Stephen Hawking
(C) Natural Language Processing thinks the primitive forms
of A.I. which have been de-
(D) none of above veloped so far have already
proved very useful.
40. For the image classification
problem, which of the fol- (A) basic
lowing neural networks is (B) modern

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 67

(C) original following questions. By in-

(D) recent venting revolutionary new
technologies, such a super
45. ‘Siri’ and ‘Alexa’ are types intelligence might help us
of eradicate war, disease and
poverty.
(A) Artificial Narrow Intelligence
(B) Artificial General Intelligence (A) erect

(C) Artificial Super Intelligence (B) illiterate

(D) Artificial Strong Intelligence (C) generate
(D) exterminate
46. In , the machine is trained
with huge amounts of data 50. Which AI technique enables
which helps it in training it- the computers to under-
self around the data. stand the associations and
(A) Supervised Learning relationships between ob-
(B) Deep Learning jects and events?

(C) Classification (A) Heuristic Processing

(D) Unsupervised Learning (B) Cognitive Science
(C) Relative Symbolism
47. Which of these jobs COULD
be taken by a robot using (D) Pattern Matching
AI?
51. Guess what virtual reality
(A) Teacher
is.
(B) Doctor
(A) Digitally created environment
(C) Nurse around the user using a display.
(D) Bank clerk (B) Putting virtual elements in the
48. A certain Professor at the real world.
Stanford University coined (C) Websites with 3D content.
the word ‘artificial intelli-
(D) What your phone camera
gence’ in 1956 at a confer-
sees.
ence held at Dartmouth col-
lege. Can you name the Pro- 52. An Al system processing
fessor? CCTV camera recordings to
(A) David Levy recognise a face is a sce-
(B) Joseph Weizenbaum nario referring to which of
the following Al domains?
(C) John McCarthy
(A) Data
(D) Hans Berliner
(B) Computer Vision
49. Indicate the word(s) CLOS-
(C) NLP
EST in meaning to the under-
lined word(s) in each of the (D) All of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
68 Chapter 2. Application of supervised learning

53. Which of the following ca- (C) A variable name can start with
pabilities is a challenge to a symbol
develop in machines? (D) A variable name can include
(A) Retain facts as knowledge special symbols

NARAYAN CHANGDER
(B) Recall the knowledge 58. AI-based microprocessor
(C) Think, analyse and apply are computing.
logic (A) 4G
(D) All of these (B) 5G

54. To cope with the bewilder- (C) NLP

ing of the real world, sci- (D) Intel
entists often ignore less rele-
59. The intelligence displayed
vant details.
by humans and other ani-
(A) complex mals is termed?
(B) complexion (A) Constance
(C) complexity (B) Ability
(D) complexions (C) Natural Intelligence

55. Who is called the Father of (D) Cognition

Artificial Intelligence? 60. The study of computer al-
(A) John McCarthy gorithms that improve au-
tomatically through experi-
(B) Herbert Simon
ence is termed?
(C) Allen Newell (A) Compilation
(D) Allan Turing (B) Formulation
56. Learning based on reward- (C) Machine Learning
punishment approach is (D) Ability
called
61. Which of the following is
(A) Deep Learning
not a type of AI?
(B) Loop Learning
(A) Reactive Machine AI
(C) Reinforcement Learning (B) Deep AI
(D) Learning with Algorithm (C) Limited Memory AI
57. Which of the following is (D) Theory of Mind AI
NOT true about variable
62. Patient illness diagnosis
names?
system is
(A) variable name must start with
(A) Information system
a letter or the underscore charac-
ter (B) Expert system
(B) Variable name cannot include (C) Database system
spaces between words (D) Computer System

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 69

63. If there are aliens out there, 68. In training an Artificial Intel-
do you think they are much ligence system, we need to
more advanced than we provide a data set with suffi-
are? ciently large amount of pho-
(A) technological tos, speeches, or texts, etc.
for the system to
(B) technologically
(A) analyse
(C) technology
(D) technologies (B) learn
(C) manage
64. Which of the below AI
projects don’t exist yet? (D) practice
(A) An AI with citizenship
69. CLOSEST in meaningA.I. al-
(B) A robot with a muscular skele-
gorithms can also help to de-
tal system?
tect faces and other features
(C) AI that can read it’s owner’s in photo sent to social net-
emotions working sites and automat-
(D) AI that develops emotions ically organize them.
over time. (A) categorize
65. Automatic Reasoning tool is (B) connect
used in
(C) recognize
(A) Personal Computers
(D) remind
(B) Microcomputers
(C) LISP Machines 70. Which of these can be done
(D) All of the above in the field of healthcare, by
using AI?
66. These data are not in the
usual database format but (A) Detection of diseases & their
have the property of being treatment
easier to process. (B) Decision making
(A) Structured (C) Medical treatment
(B) Unstructured
(D) All of these
(C) Semi-Structured
(D) none of above 71. The term Artificial Intelli-
gence was first coined by
67. Can AI work in small, dark, whom?
airless spaces?
(A) John McCarthy
(A) Well
(B) Sure! (B) Charles Babbage

(C) Not really (C) Tim Berners Lee

(D) Cannot at the moment (D) Bill Gates

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
70 Chapter 2. Application of supervised learning

72. The biggest fear is that hu- (C) John McCarthy

mans might control over (D) Allen Newell
robots.
(A) gain 77. How do we teach an AI de-
vice?
(B) keep

NARAYAN CHANGDER
(A) Give it data
(C) lose
(B) Provide it with examples
(D) take
(C) All of the above
73. Software designed to facili- (D) None of the above
tate a conversation between
a computer and human end- 78. Which one is the ‘Expert
user? Task’ for a machine?
(A) Voice response (A) Recognizing objects
(B) Walkie Talkie (B) Solving Mathematical prob-
(C) Chatbot lems
(D) Messaging (C) Medical Diagnsoning
(D) None of these
74. Consider the following two
statements:i) AI technology 79. How many benefits of AI
is the first truly safe tech- are there in our presenta-
nology of 21st century ii) tion?
AI technology provides best
(A) 1
returns at minimum risks.
Which of these statements (B) 2
is/are true? (C) 3
(A) Only ( i ) (D) 4
(B) Only ( ii )
80. Which of these does not be-
(C) Both ( i ) and ( ii ) long to reinforcement learn-
(D) Neither ( i ) nor ( ii ) ing?
(A) rewards
75. Find the word which has a
different sound in the part (B) agent
underlined. (C) environment
(A) algorithm (D) labelled input
(B) threat
81. When was the term Arti-
(C) thousand ficial Intelligence first pro-
(D) theory posed?

76. Who is known as the-Father (A) 1920’s

of AI”? (B) 1950’s
(A) Fisher Ada (C) 2000’s
(B) Alan Turing (D) 1990’s

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 71

82. Responding in speech to the 87. Weak AI system

questions asked by the user (A) mimic intelligent behaviour
refers to which of the follow-
ing Al domains? (B) fail at reasoning based tasks
(A) Data (C) cannot handle Big Data
(B) Computer Vision (D) Can understand emotions
(C) NLP 88. What is A.I. used for?
(D) Both a) and c) (A) Healthcare
83. General games involves, (B) Transportation
(A) Single-agent (C) Education
(B) Multi-agent (D) All of the Above
(C) Only a and b 89. No wonder the vacuum
(D) Neither a nor b cleaner isn’t working. You
haven’t it
84. What is the most impor-
tant difference between (A) brought-in
batch gradient descent, (B) given-in
small batch gradient de- (C) plugged-in
scent, and stochastic gradi-
ent descent? (D) pulled-in

(A) Gradient size 90. who owns the smart phone

(B) Gradient direction named 2.0
(C) Learning rate (A) bill gates
(D) Number of samples used (B) mark zukerberg
(C) steve jobs
85. The first industrial robot to
work on a general Motors (D) none of above
assembly line is
91. The ability of a computer to
(A) Roomba interact andunderstand dif-
(B) Eliza ferent human languages
(C) Unimate (A) computer vision
(D) Kismet (B) robotics
(C) Natural Language Processing
86. Nowadays, automobile
manufacturers have indus-
trial robots human work- (D) none of above
ers in factory operations.
92. Which algorithm is used in
(A) replace the Game tree to make deci-
(B) to replace sions of Win/Lose?
(C) replacing (A) Heuristic Search Algorithm
(D) replaced (B) DFS/BFS algorithm

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
72 Chapter 2. Application of supervised learning

(C) Greedy Search Algorithm 97. Which one of this ideas is

(D) Min/Max algorithm NOT a benefit of the AI on
medicine
93. Process of producing use- (A) Efficiency, accuracy and preci-
ful patterns by processing sion

NARAYAN CHANGDER
enormous amount of data is
(B) Lack of human touch-empathy
called
and emotional intelligence
(A) Data processing
(C) Saves money and have better
(B) Sensing monitoring
(C) Analytics (D) none of above
(D) Reasoning 98. Which of the following
94. An Expert System uses the is/are the composition for
knowledge of many experts AI agents?
in a particular field to (A) Program only
(A) solve problems and make de- (B) Architecture only
cisions (C) Both Program and Architec-
(B) retain knowledge ture
(C) write game playing programs (D) None of the mentioned

99. If a robot is able to change

(D) none of above its own trajectory as per the
external conditions, then
95. By the housework done, the robot is considered as
we have time to further our the
studies.
(A) Mobile
(A) taking
(B) Non-Servo
(B) to have
(C) Open Loop
(C) having
(D) Intelligent
(D) to get
100. system that mimic the
96. What is the function of an expertise of an expert is
Artificial Intelligence “Agent known as
function”?
(A) Information System
(A) Mapping of goal sequence to
(B) Computer System
an action
(C) Expert System
(B) Work without the direct inter-
ference of the people (D) Super System
(C) Mapping of precept sequence 101. Content that changes ac-
to an action cording to the needs of the
(D) Mapping of environment se- user is called
quence to an action (A) User-friendly content

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 73

(B) Adaptive content 106. AI duet is a game which

uses
(C) Intelligent content
(A) Natural language processing
(D) Streaming content
(B) Natural image processing
102. This stores information (C) Computer vision
that the computer can use
later in the program. (D) None of these

(A) Variable 107. Researchers have to

(B) Input the conclusion that your per-
sonality is affected by your
(C) Output genes.
(D) Print (A) arrived
(B) come
103. Why are some people
afraid of AI developing? (C) got
(A) Robots may take the jobs of (D) reached
humans
108. AI is incorporated into a
(B) Robots would be better than variety of different types of
humans technology. Here are some
(C) Robots haven’t got a good of the examples EXCEPT
brain (A) Automation
(D) Robots are too expensive (B) Machine vision
(C) SIRI
104. OPPOSITE in meaningDo
you really think humans (D) Natural language processing
will be exterminated when
more and more intelligent 109. Artificial Intelligence is a
machines are invented in way of
the future? (A) software think intelligently
(A) killed (B) a computer-controlled robot
(B) restored (C) making a computer
(C) resurrected (D) All of the above
(D) wiped out 110. Photo editing features
such as brightness and con-
105. NLP can be divided into trast in applications such as
these: Photoshop. Is this AI
(A) NLU and NLG (A) Yes
(B) NLP and NLC (B) No
(C) NLG and NLE (C) Kind of
(D) none of these (D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
74 Chapter 2. Application of supervised learning

111. After the power failure 116. for understanding hu-

last night, the washing ma- man emotions.
chine started to , so I
(A) Computer Vision
had to wash the clothes by
hands. (B) Natural Language Processing

NARAYAN CHANGDER
(A) make noise (C) Neural Networks
(B) malfunction (D) None of these
(C) operate
117. When the term “Artifi-
(D) work cial Intelligent” was first
coined?
112. Chat engines enabled with
Al are called (A) in 1956

(A) Robots (B) in 1946

(B) Computer Machines (C) in 1936

(C) Chatbots (D) in 1966

(D) None of these 118. Which of the following is a
positive impact of AI on soci-
113. Expert systems are part of
ety?
a general category of com-
puter applications known (A) Data privacy
as (B) Biased judgments
(A) artificial intelligence (C) Artificial intelligence working
(B) human expert hours
(C) computer integration (D) Loss of control
(D) experts
119. Artificial intelligence (AI) is
114. The idea of ‘The Imitation an area of
Game’ was proposed by (A) Electronics
(A) Rossum (B) IOT
(B) Arthur Samuel (C) Advanced physics
(C) Alan Turing (D) Computer Science
(D) John McCarathy
120. A self-driven car is an ex-
115. They have just had air con- ample of which of the fol-
ditioning in their house. lowing?
(A) install (A) Reactive machines
(B) to install (B) Theory of mind
(C) installed (C) Limited memory
(D) installing (D) Self-awareness

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 75

121. Which one of these jobs (C) f(n)<=min(g(n), h(n))

COULD NOT ethically be (D) h(n)<=c(n, a, n’)+h(n’)
taken by a robot with AI?
126. My sister her ear
(A) Doctor
pierced last weekend.
(B) Bank clerk
(A) did
(C) Telephone sales person (B) got
(D) Shop worker (C) made
122. Most scientists believe that (D) took
the day when robots be-
127. Machine Learning is a sub-
come part of our lives will
set of Artificial Intelligence
come
which allows a machine to
(A) from time to time make by learning from
(B) more or less experiences.
(C) sooner or later (A) preparation
(D) later than never (B) predictions
(C) testing
123. “Fish is a non-vegetarian
dish, Rajeev loves to (D) precaution
eat non-vegetarian dishes; 128. In this type of learning
therefore, Rajeev loves to neural networks are in-
eat fish”. This statement is volved.
an example of which rea-
(A) ML
soning:
(B) DL
(A) Inductive
(C) Supervised ML
(B) Deductive
(D) None of these
(C) None of these
129. High intelligent machines
(D) Both
can be automated to oper-
124. is the personal digital ate without human
assistant developed by Mi- (A) intervene
crosoft and works for win-
(B) intervening
dows 10.
(C) intervention
(A) google assistant
(D) interventionist
(B) cortana
130. Which of the languages
(C) apple siri
below are used to code a
(D) alexa robot?
125. Which one is correct in A* (A) Python
search (B) Machine Learning
(A) f(n)=h(n) (C) Java
(B) f(n)=g(n)+h(n) (D) Computer Vision

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
76 Chapter 2. Application of supervised learning

131. One of the following is not 136. Fill in the blanks:-In year
an example of A.I. , college the concept of
(A) ATM artificial intelligence was de-
veloped.
(B) Robot
(A) 1956, Darthmouth

NARAYAN CHANGDER
(C) Bicycle
(B) 1986, Darthmouth
(D) Self-driving car
(C) 1959, Darthmouth
132. Who is the quotee of “It
(general AI) would take off (D) 1976, Darthmouth
on its own, and re-design 137. Which of the following ar-
itself at an ever increas- eas can not contribute to
ing rate. Humans, who build an intelligent system
are limited by slow biologi-
cal evolution, couldn’t com- (A) Neuron science
pete, and would be super- (B) Maths
seded.”
(C) Computer Science
(A) Stephen Hawking
(D) Geology
(B) Barack Obama
(C) Christopher Nolan 138. For propositional Logic,
which statement is false?
(D) Bill Gates
(A) The sentences of Propositional
133. Data is the key of the logic can have answers other than
domains of Al. True or False.
(A) Two (B) Each sentence is a declarative
(B) Three sentence.
(C) Five (C) Propositional logic is a knowl-
(D) Eight edge representation technique in
AI.
134. What is Facefirst (D) None of the above
(A) Prediction software
139. Which of the following
(B) Facial recognition software
helps you in assessing ma-
employed by Walmart
chine learning algorithms?
(C) A branch of Facebook,
(A) K-Means Clustering
(D) none of above
(B) Linear Regression
135. What was the name of the (C) Data Analysis
first computer created dur-
ing WW II? (D) Performance Metrics
(A) IBM Deep Blue 140. What is artificial intelli-
(B) ENIAC gence used for?
(C) ENIGMA (A) For reading
(D) none of above (B) For writing

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 77

(C) For driving a car without a the trajectory you want it to

driver. follow is called:
(D) For nothing. (A) Robot vision control

141. Mark the letter A, B, C, (B) Continuous path control

or D to indicate the under- (C) Contact sense control
lined part that needs correc- (D) Pick and place control
tion in each of the follow-
ing sentences:Artificial intel- 146. Python was released in
ligence is also known for the year
machine intelligence.
(A) 1951
(A) artificial
(B) 1947
(B) also
(C) 1956
(C) for
(D) 1991
(D) intelligence.
147. Is Sophia Robot an exam-
142. The field that investigates ple of AI?
the mechanics of human in-
(A) Yes
telligence is:
(B) No
(A) History
(C) May be
(B) Cognitive science
(D) none of above
(C) Psychology
(D) Sociology 148. A machine that is special-
ized in one task or area.
143. keeps in memory all
(A) Artificial general intelligence
of the already generated
nodes (B) Artificial narrow intelligence
(A) RBFS (C) Artificial super intelligence
(B) A* (D) none of above
(C) SMA* 149. Artificial Intelligence is the
(D) IDA* intelligence demonstrated
by a that simulates hu-
144. Who is the founder of Mi- mans in cognition, learning
crosoft? and making decisions.
(A) Jeff Bezos (A) machine
(B) Bill Gates (B) car
(C) Elon Musk (C) printer
(D) Sundar Pichai (D) counting machine

145. Programming a robot by 150. What modern invention

physically moving it through gave machines enough data

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
78 Chapter 2. Application of supervised learning

to begin mimicking human 155. These are data already

intelligence? stored in the databases in
an ordered manners
(A) Instagram
(A) Structured
(B) Drones
(B) Unstructured

NARAYAN CHANGDER
(C) Tik Tok
(C) Semi-Structured
(D) Social Media
(D) none of above
151. Self driving cars are a type
of AI 156. The first true Al programs
had to await the of
(A) Reactive Machine AI stored-program electronic
(B) Theory of Mind digital computers.
(C) Limited Memory AI (A) arrive
(D) None of these (B) arriving
(C) arrival
152. Which AI is considered as
a threat to human beings (D) arriver

(A) Artificial Super Intelligence 157. Mark the letter A, B, C,

or D to indicate the word
(B) Artificial General Intelligence
whose underlined part dif-
(C) Artificial Narrow Intelligence fers from the other three in
(D) None of these pronunciation in each of the
following questions.
153. is a domain of AI that (A) automated
depicts the capability of a
(B) controlled
machine to get and analyse
visual information and af- (C) invented
terwards predict some deci- (D) substituted
sions about it.
158. OPPOSITE in meaningS-
(A) NLP
mart robots have replaced
(B) Data Sciences humans in stressful and haz-
(C) Augmented Reality ardous jobs and in assem-
bly lines.
(D) Computer Vision
(A) dangerous
154. Which of the following is (B) risky
a properly written print com-
(C) safe
mand?
(D) perilous
(A) print(”Hello World!”)
(B) print(Hello World!) 159. earliest successful Al
program was written in
(C) print”Hello World!”
1951 by Christopher Stra-
(D) prnt(”Hello World!”) chey, later director of

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 79

the Programming Research (C) Playing a game

Group at the University of
(D) Making a Machine Intelligent
Oxford.
(A) A 164. Artificial intelligence is all
(B) An about robots.
(C) The (A) True
(D) That (B) False
160. Which technology enables (C) Maybe
features such as speech
recognition, product recom- (D) none of above
mendations, and credit card
fraud detection? 165. FM is a method of
(A) natural intelligence signals, especially in radio
broadcasting.
(B) machine learning
(A) transacting
(C) IBN
(D) artificial intelligence (B) transferring
(C) transmitting
161. Which of these jobs COULD
easily be taken by a robot (D) translating
using AI?
(A) Teacher 166. One of the popular
used in smartphones at
(B) Doctor
present is voice recogni-
(C) Nurse tion.
(D) Bank clerk (A) applicant
162. It is a software that ex- (B) applicable
hibits analytical, decision
making and learning abili- (C) applications
ties similar to those of hu- (D) apply
mans.
(A) Artificial Intelligence 167. Treatment chosen by doc-
(B) Artificially Intelligence tor for a patient for a dis-
ease is based on
(C) Artificial Intelligent
(D) Artificially Intelligent (A) Only current symptoms
(B) Current symptoms plus some
163. What is Artificial intelli-
knowledge from the textbooks
gence?
(A) Programming with your own (C) Current symptoms plus some
intelligence knowledge from the textbooks
plus experience
(B) Putting your own intelligence
into computer (D) Only a and b

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
80 Chapter 2. Application of supervised learning

168. Artificial Intelligence is 173. have announced that

about a major breakthrough has
been made.
(A) Playing a game on Computer
(A) Research
(B) Researchers

NARAYAN CHANGDER
(B) Programming on Machine
with your Own Intelligence (C) Researches
(C) Making a machine Intelligent (D) Researching
(D) Putting your intelligence in 174. The old astronomer pa-
Machine tiently made his and
wrote down what he saw.
169. Who was the funder of Ar-
tificial Intelligence (A) observation
(B) observatory
(A) John McCarthy
(C) observe
(B) Christine Gabbard
(D) observer
(C) Ona Juodkiene
175. These programs act intelli-
(D) Sara Hines
gently but are not intelligent.
170. GPS systems help a lot of They are
people on land, on the (A) Weak AI
sea, and in the air. (B) Strong AI
(A) inform (C) Not Sure which category
(B) interfere (D) None of these
(C) navigate 176. CLOSEST in meaningInter-
(D) reunite net search engines can help
to predict the weather as
171. The first general purpose well as traffic jams in a city.
mobile robot, also called the (A) announce
‘first electronic person’ is
(B) expect
(A) David
(C) forecast
(B) ASIMO
(D) reveal
(C) Unimate
177. Which part of this pro-
(D) Shakey gram is the variable?
robot name = “Wall-
172. An agent is composed of
E”print(”The cutest robot is”
+ robot name)
(A) Architecture (A) robot name
(B) Agent Function (B) “Wall-E”
(C) Perception Sequence (C) “The cutest robot is”
(D) Architecture and Program (D) print

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 81

178. A theoretical type of AI (B) Robotics

where the machine can out-
(C) Cybernetics
perform human intelligence
in every task (D) none of above
(A) Artificial general intelligence 183. First order logic State-
(B) Artificial narrow intelligence ments contains
(C) Artificial super intelligence (A) Predicate and Preposition
(D) none of above (B) Subject and an Object
179. Which of the following (C) Predicate and Subject
have people traditionally (D) None of the above
done better than comput-
ers? 184. ANI stands for:
(A) Recognising relative impor- (A) Artificial Narrow Intelligence
tance
(B) Artificial Normal Intelligence
(B) Finding similarities
(C) Artificial Number Intelligence
(C) Resolving ambiguity
(D) Artificial Neuron Intelligence
(D) All of the above

180. the artificial intelligence 185. What is a computer sys-

timeline, the period from tem that emulates the
1974-93 was known as decision-making ability of a
human expert.
(A) AI winter
(A) Decision system
(B) AI summer
(B) Expert System
(C) AI cold
(C) Experiment System
(D) AI hot
(D) All of the above
181. Which company built a
computer that took 40 min- 186. Self-driving car is an appli-
utes to carry a single second cation of computer
of neural activity
(A) regocnition
(A) Fujitsu
(B) data
(B) Microsoft
(C) vision
(C) IBM
(D) task
(D) Commodore 64

182. What science is responsi- 187. In AI; CV stands for:

ble for studying the mech- (A) Computer Version
anisms of communication
(B) Computer Vision
and control on both ma-
chines and humans? (C) Computer Vote
(A) Informatics (D) None of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
82 Chapter 2. Application of supervised learning

188. Stimulation of human in- 193. Which of the following is

telligence which is pro- NOT an example of artificial
cessed by machines is called narrow intelligence?
(A) Google Translate
(A) Artificial Intelligence

NARAYAN CHANGDER
(B) Super-intelligent AI agents
(B) Vision Systems
(C) Spam Filter
(C) Natural Language
(D) AlphaGo
(D) none of above

189. The process of imitating 194. Communication means

the working of a human (A) To share
brain for processing data
and used to make decisions (B) To prove
(A) Robotics (C) To answer
(B) Machine Learning (D) To give
(C) Deep learning
195. UFO is an acronym for
(D) Decision Support Flying Object.
190. What year do futurists be- (A) Unidentified
lieve A.I will meet adult hu-
man intelligence? (B) Unimaginative
(A) 2025 (C) Unknown
(B) 2035 (D) Unique
(C) 2040
196. Who is considered to be
(D) 2029 the “father” of artificial in-
telligence?
191. Which process makes two
different Logical expres- (A) John McCarthy
sions look identical?
(B) Alan Ada
(A) Unification
(C) John border
(B) Lifting
(D) Charlie Chaplin
(C) Inference Process
(D) None of the above 197. If a robot can alter its own
trajectory in response to ex-
192. Which of these is a tool
ternal conditions, it is consid-
used in Artificial Intelli-
ered to be:
gence?
(A) Intelligent
(A) Art
(B) Input (B) Mobile
(C) Design (C) Servo
(D) Neural Networks (D) None

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 83

198. A.I. robots are of hav- 203. Who is known as the “Fa-
ing human emotions. ther of AI “?
(A) able (A) Fisher Ada
(B) capable (B) Alan Turing
(C) efficient (C) John McCarthy
(D) fond (D) Allen Newell
199. Which of the following 204. Which of the following
does not belong to the stage was a chess playing super
‘ Data Exploration’? computer made by IBM?
(A) Gathering Data (A) Blue coral
(B) Data Analysis (B) Deep Blue
(C) Data Preparation (C) Watson
(D) Data Wrangling (D) Sophia
200. states that about ev-
205. Which of the following is
ery 2 years, technology ad-
a fundamental goal of re-
vances so that the speed of
search in Artificial Intelli-
computers doubles.
gence?
(A) Moore’s Law
(A) Reasoning
(B) Trump’s Law
(B) Coupling
(C) Murphy’s Law
(C) Mastering
(D) Lawrence’s Law
(D) Data
201. The application/applications
of Artificial Intelligence 206. AI analyzes more and
is/are deeper data using

(A) Expert Systems (A) Computer Planning

(B) Gaming (B) Neural Networks

(C) Vision Systems (C) Computer science

(D) All of the above (D) none of above

202. What is an algorithm? 207. This AI can be used to

check patients into hospitals
(A) A very difficult physique prob-
and collect patient informa-
lem.
tion.
(B) AN app designed to solve
(A) Data and statistics
maths homework.
(B) AI-assisted robotic surgery
(C) A set of instructions designed
to perform a specific task. (C) Virtual nursing assistants
(D) none of above (D) Image analysis

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
84 Chapter 2. Application of supervised learning

208. Machines with strong intel- 213. What was the name of
ligence are able to retain the computer that defeated
(A) Data Wold Champion Gary Kas-
parov in 1997?
(B) Learning
(A) Deep Sea

NARAYAN CHANGDER
(C) Language
(B) Ocean Blue
(D) Skill
(C) Old Sea
209. The of the outbreak (D) Deep Blue
of the disease is still un-
known. 214. which AI system passed
an eight grade science test
(A) aim
developed by the allen insti-
(B) cause tute?
(C) motive (A) deep blue
(D) reason (B) deep learning
210. What is artificial intelli- (C) aristo
gence said to be? (D) sophiya
(A) very dangerous
215. What is AI used for?
(B) intelligent robots
(A) Healthcare
(C) smarter than humans
(B) Transportation
(D) the future
(C) Education
211. Which of the following (D) All of the Above
tasks would an AI be able to
perform the LEAST success- 216. The meaning of artificial
fully? intelligence was first pro-
posed by a scientist in 1950,
(A) Suggesting what someone
and at thesame time a test
might want to buy from a shop-
model of machine intelli-
ping website
gence was proposed. Who
(B) Using traffic data to work out is this scientist?
what cars people are likely to buy
(A) Minsky
(C) Playing hangman
(B) Zade
(D) Writing a joke
(C) Turing
212. WHAT ARE AI’S FEEDED UP (D) Von Neumann
WITH?
217. The following are exam-
(A) QUALITY DATA ples of an expert system ex-
(B) JUNK FOOD cept
(C) COMMON SENSE (A) Automated Teller Machine
(D) none of above (B) Thermometer

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 85

(C) Traffic Light 223. The main tasks of an AI

(D) System unit agent are
(A) Input and Output
218. Among the given options,
which search algorithm re- (B) Moment and Humanly Actions
quires less memory?
(A) Optimal Search (C) Perceiving, thinking, and act-
(B) Depth First Search ing on the environment
(C) Breadth-First Search (D) None of the above
(D) Linear Search
224. Which of these is not a
219. A humanoid robot known Google Product?
as the first “Robot Citizen” is (A) Pixel
called
(B) Echo
(A) Sophia
(B) Clarke (C) Nest

(C) Roomba (D) Waymo

(D) Eliza 225. Categorize Crossword
220. When the term “AI” first puzzle in Fully Observable
coined? / Partially Observable.
(A) Dartmouth Conference (A) Fully Observable
(B) SAARC meeting (B) Partially Observable
(C) UNICEF meeting (C) All of the mentioned
(D) None of these (D) None of the mentioned
221. What is the field of com-
puter science responsible for 226. Which one can be consid-
managing large amounts of ered an application of AI?
data? (A) Siri
(A) Machine Learning (B) WhatsApp
(B) Big data (C) Email
(C) Deep learning
(D) none of above
(D) none of above
227. In which year do you think
222. How does an AI device
the term ‘Artificial Intelli-
learn and think?
gence’ was coined?
(A) It hacks our computer
(A) 1950
(B) It accepts data and trains itself,
then creates a model (B) 1960
(C) It hacks our brain (C) 1966
(D) None of the above (D) 1956

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
86 Chapter 2. Application of supervised learning

228. What can AI do? 233. Alexa is a type of

(A) Can show emotions (A) chatter box
(B) Ability to self-perceive your- (B) chatbox
self, consciousness and behave (C) chatbot

NARAYAN CHANGDER
like people
(D) chat box
(C) Understand human emotions
234. The area of A.I that inves-
(D) All are right tigates methods of facilitat-
ing communication between
229. Which of the following is people and computers:
not a type of AI based on (A) Artificial language processing
functionality?
(A) Reactive Machines (B) Symbolic processing
(B) Theory Of Mind (C) Natural Language processing
(C) Self Minded
(D) Limited Memory (D) Decision support

230. Many materials have 235. I don’t think people should

been used for teeth, in- be allowed to experi-
cluding wood. ments on animals.
(A) artificial (A) create
(B) false (B) make
(C) hand-made (C) perform
(D) natural (D) produce

231. Which is the last stage of 236. capability of Al helps in

AI Project Cycle? analysis of visual data such
as images.
(A) Evaluation
(A) Computer vision
(B) Deployment
(B) Machine learning
(C) Problem Scoping
(C) NLP
(D) Testing
(D) All of these
232. Spreadsheet that calcu-
lates sums and other pre- 237. Which of the basic parts
defined functions on given of a robot unit would in-
data. Is this AI clude the computer circuitry
that could be programmed
(A) Yes to determine what the robot
(B) No would do?
(C) Kind of (A) controller
(D) none of above (B) end effector

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 87

(C) sensor 243. An AI agent perceives and

(D) drive acts upon the environment
using
238. We had the roof last (A) Sensors
year.
(B) Perceiver
(A) mend
(C) Actuators
(B) mending
(D) Both a and c
(C) mended
(D) being mended 244. the bulk of data, more
accurate are the predictions
239. What is a disadvantage of by AI.
AI? (A) Organised
(A) Someone always needs to con- (B) Lower
trol it
(C) Higher
(B) It’s affordable
(D) Structured
(C) It costs a lot of money
(D) All of above 245. Expert systems have been
developed to mimic the
240. Select advantages of A.I. of an expert in a particular
field.
(A) Common sense
(A) both
(B) Less error
(B) expertise
(C) No creativity
(C) knowledge
(D) Human touch
(D) none of above
241. What are scientists trying
to replicate when develop- 246. The creator of a computer
ing AI? language called ‘LISP’ is
(A) The human brain (A) John McCarathy
(B) The human heart (B) Herbert Simon
(C) The human mouth (C) Allen Newell
(D) The human robot (D) Allan Turing

242. Which is used to improve 247. Do you know who the

the performance of heuristic fact that sound travels in
search? waves?
(A) Keeps nodes as Simple form (A) developed
(B) Quality of nodes (B) discovered
(C) Quality of heuristic function (C) found
(D) None of the mentioned (D) invented

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
88 Chapter 2. Application of supervised learning

248. Artificial intelligence is (B) Natural Language Under-

? standing
(A) Not real (C) National Language Under-
standing
(B) Playing a Game
(D) None of the Above

NARAYAN CHANGDER
(C) Making a Machine intelligent
(D) none of above 254. These data have no clear
format in storage and not in
249. Among the machine learn- the traditional database for-
ing algorithms, which of the mat.
following are not integrated (A) Structured
learning strategies?
(B) Unstructured
(A) Boosting
(C) Semi-Structured
(B) Stacking
(D) none of above
(C) Bagging
255. Choose the correct option
(D) Marking
for the following statement:-
250. is the ability of a com- A) Knowledge base (KB) is
puter to perform tasks that consists of set of statements.
require human intelligence B) Inference is deriving a
new sentence from the KB.
(A) skill
(A) A is false, B is true
(B) AI
(B) A is false, B is false
(C) Data
(C) A is true, B is false
(D) Application
(D) A is true, B is true
251. You should have your car
256. Now gardeners find their
before going.
job easier by getting the
(A) servicing grass by automatic lawn
(B) to service mowers.
(C) service (A) cuts
(D) serviced (B) to cut
(C) cutting
252. Who is the father of AI?
(D) cut
(A) Alan Turing
257. Which of these does not
(B) Albert Einstein
use Machine Learning / AI?
(C) John McCarthy
(A) driverless cars
(D) Issac Newton
(B) Siri / Alexa
253. NLU Stands for (C) Wireless speakers
(A) Neutral Language Under- (D) facial recognition on your
standing phone

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 89

258. Which stage of AI are peo- (C) Python

ple particularly concerned
(D) Cobol
about?
(A) Narrow AI 263. The purpose of AI is to
(B) Human level AI (A) Build problems, learns and
(C) Super AI understand language.

(D) none of above (B) Build emotions, language and

understands mathematical prob-
259. OPPOSITE in mean- lems.
ingKurzweil believes that (C) Build computer programs that
there could be tiny robots exhibit intelligent behavior
called nanobots implanted
into our brains to improve (D) none of above
our memory.
264. A mistake in code that
(A) inserted in causes the computer to do
(B) fixed in something different than
what you expected it to do.
(C) removed from
(A) Bug
(D) replaced with
(B) Syntax
260. An AIO system trying to
recognise a snapshot of a (C) Output
handwritten note. This inte- (D) Variable
grates which of the follow-
ing domains of Al? 265. A brand new AI program
is similar to making a
(A) Data
guess.
(B) Computer Vision
(A) a teacher
(C) NLP
(B) a dog
(D) All of these
(C) an adult
261. It is an artificial chatting
(D) a baby
robot or chat box that re-
sponds to your questions. 266. Mark the letter A, B, C, or
(A) SimiSimi D to indicate the word that
differs from the other three
(B) SimSimi
in the position of primary
(C) SimSimmi stress in each of the follow-
(D) SimmSimi ing questions.
(A) intelligence
262. which of these languages
is used for Ai programming (B) diversity
(A) C++ (C) automated
(B) Fortran (D) exterminate

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
90 Chapter 2. Application of supervised learning

267. The three stages of artifi- 272. Weak artificial intelligent

cial intelligence include com- systems:
putational intelligence, , (A) show intelligent behaviour
and cognitive intelligence:
(B) lack natural language pro-
(A) Weak AI cessing skills

NARAYAN CHANGDER
(B) Perceptual Intelligence (C) cannot perform complex tasks
(C) Behavioral Intelligence
(D) Strong AI (D) can achieve self conscious-
ness with time
268. He was barely of writ-
273. An expert system is com-
ing his own name.
posed of a
(A) able
(A) Database, query engine
(B) capable
(B) Knowledge base, search en-
(C) enable gine
(D) talented (C) Database, inference engine
(D) Knowledge base, inference
269. An interactive computer
engine
program that could function-
ally converse in English with 274. The AI application that
a person is transports you into a three-
(A) Kismet dimensional world is
(B) Shakey (A) Virtual learning

(C) Eliza (B) Virtual reality

(C) Virtual Intelligence
(D) Unimate
(D) none of above
270. The unique characteristics
of something 275. Which agent deals with
the happy and unhappy
(A) Supervised learning state?
(B) Features (A) Utility-based agent
(C) Unsupervised learning (B) Model-based agent
(D) none of above (C) Goal-based Agent
271. The system that allows a (D) Learning Agent
computer to make decisions, 276. Which rule is applied for
such as in health care and fi- the Simple reflex agent?
nancial advice is called
(A) Simple-action rule
(A) Experiment system
(B) Simple &Condition-action rule
(B) Expert system
(C) Random search generator (C) Condition-action rule
(D) none of above (D) None of the above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 91

277. A technique that was (A) Smart assistants

developed to determine (B) Email filters
whether a machine could or
could not demonstrate the (C) Self driving cars
artificial intelligence known (D) Price comparison websites
as the
(A) Boolean Algebra 282. In the operation of com-
puter, we need to tell a com-
(B) Logarithm puter how to solve a prob-
(C) Turing Test lem by designing the de-
tailed steps in an
(D) Algorithm
(A) calculation
278. User’s regular brows-
ing habits together make (B) counting
his/her (C) algorithm
(A) Browser fingerprinting (D) mathematics
(B) Browsing signature
283. The name of the female
(C) Both a) and b) intelligent robot created by
(D) None of these Hanson Robotics company
is
279. The performance of an (A) Wall-E
agent can be improved by
(B) Sophia
(A) Perceiving (C) Roomba
(B) Learning (D) none of above
(C) Observing 284. Which of these does NOT
(D) None of the mentioned use machine learning/AI?

280. Which of the following (A) driverless cars

types of tasks are compara- (B) SIRI/Alexa
tively easy for artificial intel- (C) Sonos wireless speakers
ligent systems? (i) Common
tasks(ii) Expert tasks (D) facial recognition on your
phone
(A) (i) are easier
(B) (ii) are easier 285. Which of the follow-
ing technology is used by
(C) both (1) and (ii) are equally
games like Emoji Scavenger
easy
Hunt?
(D) both (i) and (ii) are equally dif-
(A) Natural Language Processing
ficult
(B) Natural Image Processing
281. Which of the following is
an example of computer vi- (C) Computer vision
sion. (D) Neural Data Images

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
92 Chapter 2. Application of supervised learning

286. Which of these does not (A) Robotics

depict AI? (B) Natural Language Processing
(A) Autonomous Car (C) Computer Vision
(B) Notes counting machine (D) none of above

NARAYAN CHANGDER
(C) Google Home 292. Style transfer filters in ap-
(D) code that can play chess plications such as Prisma
that take a photo and trans-
287. We had the computer tech- form it into different art
nician the new software styles (impressionist, cubist,
for us. ). Is this AI
(A) to install (A) Yes
(B) installed (B) No
(C) installing (C) Kind of
(D) install (D) none of above

288. The term ‘AI’ was coined 293. Machine Learning enables
by a machine to “recognize”
and “learn” the patterns in
(A) Ken Jennings
the training set of data. The
(B) Alan Turing machine learns by looking
(C) John McCarthy for patterns from the train-
ing data set and then builds
(D) Brad Rutter
a
289. is a subset of AI which (A) pattern
enables machines to im- (B) module
prove at taskes with experi-
(C) machine
ence (data).
(D) model
(A) Machine learning
(B) Computer vision(CV) 294. What is an artificial intelli-
gence robot?
(C) NLP
(A) A robot that learns by observ-
(D) Deep learning ing people
290. How many types of AI is (B) A robot that welds things
there (C) A robot that makes things
(A) 2 (D) A robot that is in the army
(B) 6 295. Which of these is not an AI
(C) 10 program
(D) 4 (A) Quick, Draw!
(B) Chatter-bot
291. The ability of a computer
to understand images and (C) Quizizz
videos (D) Spotify

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 93

296. A is divided into multi- 301. I’ll have the boys the
ple layers and each layer is chairs.
further divided into several (A) paint
blocks called nodes.
(B) to paint
(A) Neural Networks
(C) painted
(B) Convolutional Neural Net-
work (CNN) (D) painting
(C) Machine learning algorithm 302. In Artificial Intelligence,
(D) Hidden Layers NLP stands for

297. The Al system in existence (A) Natural Learning Processing

today cannot: (B) Natural Location Processing
(A) understand newspaper arti- (C) Natural Logic Processing
cles and human speeches. (D) Natural Language Processing
(B) do translations.
303. What do GP and ES stand
(C) predict human behaviour.
for?
(D) recognise objects in images.
(A) Genetic programming and Ex-
298. I can learn and under- pert system
stand like humans (B) Generic programming and Ex-
(A) Narrow AI isting systems
(B) General AI (C) General programming and
extinct sources
(C) Super AI
(D) Ghost protocol and Electronic
(D) none of above
system.
299. Which search uses the
problem specific knowledge 304. Not an example of Weak
beyond the definition of the Artificial Intelligence?
problem? (A) Google search
(A) Informed search (B) Siri, Alexa
(B) Depth-first search (C) ROSS
(C) Uninformed search (D) Self-driving cars
(D) Breadth-first search
305. The set of rules that must
300. is a modern method of be followed for the com-
building, training, and using puter to understand the in-
neural networks. structions it is given.
(A) Artificial intelligence (A) Syntax
(B) Machine learning (B) Character
(C) Neural Networks (C) Command
(D) Deep learning (D) Program

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
94 Chapter 2. Application of supervised learning

306. Face recognition is a part 311. Which form belongs to

of which AI domain? Passive Causative?
(A) CV (A) S + have + Sb + Vo + sth .
(B) NLP (B) S + have/get+ sth + Pii
(Ved/C3) (+ by Sb)

NARAYAN CHANGDER
(C) Data
(C) S+ get + Sb + To Vo+ sth
(D) All of these
(D) S + want + Sb + to Vo + sth.
307. Which of the following is
312. Ability of Al to understand
not an example of Artificial
human speech is called
Intelligence?
(A) Language processing
(A) Chess Playing Computers
(B) Human language processing
(B) Self Driving Cars
(C) Natural language processing
(C) Face Recognising Devices
(D) Human speech processing
(D) Money Counting Machines
313. What is intelligence?
308. Which is a not commonly (A) Storing or keeping lots of in-
used programming lan- formation
guage for AI?
(B) Understanding
(A) Python
(C) Being emotional
(B) R
(D) Being heartless
(C) Perl
314. What was originally
(D) Java called the “imitation game”
by its creator?
309. Who was John Mc-
Carthy? (A) Turing test
(A) The man who coined the term (B) Cybernetics
“artificial intelligence”. (C) PubG
(B) An astronaut who created a (D) None
robot.
315. Which is the third most
(C) A singer from the 80s. used AI?
(D) none of above (A) Google Assistant

310. I’ll you you to our re- (B) Amazon Alexa

search department. Please (C) Samsung Bixby
hold on. (D) none of above
(A) put-away
316. Face Lock in Smartphones
(B) put-out is an example of
(C) put-through (A) NLP
(D) put-up (B) Computer Vision

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 95

(C) ML-Data science (C) Learning theory

(D) none of above (D) Utility functions

317. Installing is firewall is one 322. Do you think Machine

method of trying to protect learning and AI are taught
against like a child.A child knows
(A) hackers only what it is taught.

(B) thieves (A) Yes

(C) robbers (B) No

(D) crooks (C) Maybe

(D) none of above
318. Expand CBT
(A) Computer Behaved Training 323. What year was Siri in-
vented?
(B) Cognitive Behavioural Ther-
apy (A) 2011
(C) Consolidated Batch of train- (B) 2000
ers (C) 2005
(D) Combined Basic Training (D) 2010
319. Which of the following is 324. Which of these is not a
believed to be of the same type of machine learning?
capacity as that of human in-
(A) Superficial Learning
telligence?
(B) Supervised Learning
(A) SuperIntelligence
(C) Unsupervised Learning
(B) Weak AI
(D) Reinforcement Learning
(C) Strong AI
(D) General AI 325. which algorithm have the
cutoff information as f-cost
320. Semantris is based on (g+h)
which of the domain of AI?
(A) A*
(A) Computer Vision(CV)
(B) IDA*
(B) Data for AI
(C) RBFS
(C) Natural Language Process-
(D) SMA*
ing(NLP)
(D) None 326. Alexa is an example of
domain
321. The action of the Simple
reflex agent completely de- (A) Natural Language Processing
pends upon (B) Machine Learning
(A) Perception history (C) Computer Vision
(B) Current perception (D) none of above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
96 Chapter 2. Application of supervised learning

327. Which one the following is (B) “1945”

not a domain of AI?
(C) print(1945)
(A) Computer Vision
(D) error
(B) Data science

NARAYAN CHANGDER
(C) Machine learning 333. Understand
(D) NLP
(A) Segment
328. The ability of a computer (B) Respectively
to understand human lan-
guage is called (C) Comprehend
(A) NLP (D) Accuracy
(B) Computer Vision
334. Kelly wanted a live band
(C) Medical Science
at her wedding.
(D) none of above
(A) been playing
329. NASA has had an au-
tonomous spaceship the (B) to play
universe recently. (C) played
(A) to explore
(D) to be played
(B) explored
(C) explore 335. Astronomers have a
new planet outside our
(D) exploring
galaxy.
330. Who is the father of artifi- (A) invented
cial intelligence?
(A) Alan Turing (B) unearthed

(B) John McCarthy (C) developed

(C) George.F.Luger (D) discovered
(D) Lady Ada Lovelace
336. Why haven’t scientists
331. Which one of these jobs been completely successful
COULD NOT be taken by a when developing AI?
robot with AI?
(A) They do not have the time or
(A) Doctor money
(B) Bank clerk
(B) The human brain is very sim-
(C) Telephone sales person ple
(D) Shop worker
(C) The human brain is very com-
332. What would be the out- plex
put? print(1945) (D) They are trying to replicate the
(A) 1945 human heart

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 97

337. Natural language process- 342. Which of the following is a

ing is . negative impact of AI on so-
(A) Using human language to ciety?
communicate with machines. (A) Artificial intelligence working
(B) Using machine language to hours
communicate with humans. (B) Data privacy
(C) Coding in natural human lan-
(C) Not subjective
guage
(D) Perform repetitive tasks with
(D) Using natural language to
ease
talk to other humans

338. There are Sustainable 343. Ways to achieve AI in real-

Development Goals life are
(A) 17 (A) Machine Learning
(B) 19 (B) Deep Learning
(C) 20 (C) Both a & b
(D) 27
(D) None of the above
339. After applying conditional
Probability to a given prob- 344. What makes humans intel-
lem, we get ligent is their ability to
(A) 100% accurate result (A) Sense
(B) Estimated Values (B) Think
(C) Wrong Values (C) Reason
(D) None of the above (D) Read
340. Which of the following is
345. A.I. stands for?
not a stage of AI?
(A) Artificial Narrow Intelligence (A) Artificially Intelligent
(B) Artificial Natural Intelligence (B) Artificial Intelligence
(C) Artificial Super Intelligence (C) Artificial Intelligent
(D) Artificial General Intelligence (D) Artificial Intelligence
341. Once they’ve finished the
346. Products you already use
extension to the , there’ll
can be improved enor-
be room for them to take on
mously with
a lot more workers.
(A) Better components
(A) factory
(B) industry (B) Best aesthetic
(C) manufacturing (C) AI capabilities
(D) trade (D) Hardware Improvements

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
98 Chapter 2. Application of supervised learning

347. There are types of Arti- 352. Computer intelligence in

ficial intelligence one specific area is called:
(A) 1 (A) Narrow AI
(B) 2 (B) General/Human Level AI

NARAYAN CHANGDER
(C) 3 (C) Super AI
(D) 4 (D) none of above

348. What is A.I.? 353. To create a comment in

Python, what symbol do
(A) When robots shut down auto- you use?
matically
(A) #
(B) Technology that performs
tasks that normally a human (B) ” “
would (C) ( )
(C) Discovering technology for (D) *
the first time
354. An example of ‘machine
(D) None of the above learning’ is
349. He has succeeded in his (A) voicemail messages of your
career in television. favorite celebrities
(A) resurrect (B) installing a new hard drive on
a computer
(B) to resurrecting
(C) a calculator that can do hard
(C) resurrecting math problems
(D) resurrected (D) a machine that improves per-
formance based on watching a
350. Which of the following
human do the same task
statements is true?
(A) Higher the bulk of data, more 355. AI learns and discovers
accurate is the prediction through
(B) Lower the bulk of data, more (A) Experiences
accurate is the prediction. (B) Data
(C) None of these (C) Speech
(D) none of above (D) Touch

351. I wish you would the 356. IBM created a question-

TV and go outside and get answering computer that
some exercise. defeated television game
show “Jeopardy!”’ champi-
(A) call off
ons in 2011. What was its
(B) give off name?
(C) put off (A) Watson
(D) turn off (B) Jenny

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 99

(C) Sherlock (B) 3 Levels

(D) Oliver (C) 4 Levels
357. Scientists are trying to (D) None of the above
out the ways to reduce pol-
lution from aircraft. 362. It is likely that artificial in-
telligence might decide to
(A) break an end to humanity simply
(B) call because it surpasses human
intelligence.
(C) turn
(D) work (A) come
(B) make
358. Strong AI system
(C) put
(A) are able to defeat human be-
ings in combat (D) take
(B) are able to mimic intelligence 363. Which of the following ar-
without being intelligent tificial intelligence systems
(C) are capable of true intelli- are most common today?
gence (A) Artificial Narrow Intelligence
(D) are unable to undertake com- Systems
mon tasks
(B) Artificial General Intelligence
359. The practice of learning Systems
from observation and mak- (C) Strong General Intelligence
ing predictions about some- Systems
thing:
(D) Strong Narrow Intelligence
(A) Neural Network Systems
(B) Machine Learning
364. Which of the following is
(C) Natural Language Processing NOT a goal or feature of AI?
(A) Forward planning
(D) Deep Learning
(B) Displaying emotion
360. Expert Ease was devel-
(C) Reasoning/decision making
oped under the direction of
(A) John McCarthy (D) none of above

(B) Donald Michie 365. The canvas helps you in

(C) Danny identifying the key elements
related to the problem.
(D) Alan Turing
(A) Problem scoping
361. A knowledge-based agent
(B) 4Ws Problem
can be defined with lev-
els. (C) Project cycle
(A) 2 Levels (D) Algorithm

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
100 Chapter 2. Application of supervised learning

366. Mark the letter A, B, C, or 370. What perceives it’s envi-

D to indicate the underlined ronment by sensors and acts
part that needs correction in upon an environment by ef-
each of the following sen- fectors?
tences:There are a number (A) Robot

NARAYAN CHANGDER
of different forms of learn-
ing as applied in artificial in- (B) A.I
telligence. (C) Agent
(A) a (D) None of the above
(B) forms 371. Intelligent machines that
(C) as can learn any tasks that hu-
mans can do.
(D) in
(A) Artificial general intelligence
367. Which of the following ar- (B) Artificial narrow intelligence
tificial intelligence systems
will theoretically surpass hu- (C) Artificial super intelligence
man capabilities? (D) none of above
(A) Artificial Board Intelligence 372. The maximum depth to
Systems which the alpha-beta prun-
(B) Human Intelligence based Ar- ing can be applied.
tificial systems (A) Eight states
(C) Super Computing Artificial In- (B) Six states
telligent systems
(C) Ten states
(D) Artificial Super Intelligence
Systems (D) Any depth

373. Which of these questions

368. A computer can play chess,
would “Siri” or “Alexa” not
What type of learning is ex-
be able to answer?
hibited by the computer pro-
gram? (A) Is there any traffic on the way
to work?
(A) Supervised Learning
(B) What’s the weather like out-
(B) Unsupervised Learning
side?
(C) Reinforcement Learning (C) What am I having for my tea
(D) Deep Learning tonight
(D) What day is Christmas Day on
369. One of the best advan-
this year?
tages of AI is:
(A) Durability 374. In the classic convolutional
neural network model,
(B) Ease of programming
what hidden layer does the
(C) Reparability Softmax function follow?
(D) Training facility (A) Convolutional layer

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 101

(B) Pooling layer (B) Images

(C) Fully connected layer (C) Audio
(D) All of the above (D) All of these
375. What is the full form of AI 380. An expert system is com-
(A) Artificial intelligence posed of a and an
(B) Advanced intelligence (A) Database, query engine
(C) Artificially intelligence (B) Knowledge base, search en-
(D) None of the following gine
(C) Database, inference engine
376. Handwriting recognition is
related to the technology (D) Knowledge base, inference
used in engine
(A) vision systems.
381. Human face recognition:
(B) improving systems.
(C) processing systems. (A) computer vision
(D) none of these (B) robotics
377. Information that comput- (C) Natural Language Processing
ers can understand to help
make predictions and infer- (D) none of above
ences about the world.
(A) Data 382. Today’s AI systems are not
capable of:
(B) Program
(C) Algorithm (A) Interpreting visuals (images
and videos)
(D) String
(B) partially predicting human be-
378. What would the output haviour
be of this code? robot
(C) partially understanding hu-
name = “R2-D2”print(robot
man speech
name + “ is the best
robot!”) (D) recognizing faces in images
(A) R2-D2 is the best robot! 383. What symbol do we use
(B) Error between words when a vari-
(C) robot name is the best able name is made up of
robot more than one word?

(D) “R2-D2 is the best robot!” (A) Underscore ( )

(B) Space ( )
379. Al is capable to process
enormous amount of (C) Hashtag (#)
(A) Text (D) Dash (-)

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
102 Chapter 2. Application of supervised learning

384. Which of the following is 389. Which of the following

not a specific technology of tasks is not normally re-
artificial intelligence? garded as an Expert Task?
(A) Knowledge map (A) Medical diagnosis of illnesses

NARAYAN CHANGDER
(B) Machine translation
(C) Riemann geometry (B) Communication in natural
language
(D) Semantic understanding
(C) Playing complex games, like
385. A type of data that is a chess
sequence of characters and
can include letters or num- (D) Solving complex mathemati-
bers. MUST BE WITHIN QUO- cal models or problems
TATION MARKS!
390. If computers become as
(A) String smart as humans, ?
(B) Integer (A) would they do our jobs better
(C) Float than we can
(D) Data (B) will they do our jobs better
than we can
386. Which of the following is
not part of the AI Project Cy- (C) would they do our jobs better
cle? than we could
(A) Data Exploration (D) will they do our jobs better
than we could
(B) Modelling
(C) Testing 391. Which AI is also known as
(D) Problem Scoping ‘Strong AI’
(A) Artificial Super Intelligence
387. Sometimes scientists talk
about the third form of AI (B) Artificial General Intelligence
that is:
(C) Artificial Narrow Intelligence
(A) Weak AI
(D) Artificial Weak Intelligence
(B) Strong AI
(C) Superintelligence 392. Robots may be equipped
the equivalent of human
(D) None of these senses such as vision, touch,
and the ability to sense tem-
388. Name any 2 methods of
perature.
collecting data.
(A) Surveys and Interviews (A) by

(B) Rumors and Myths (B) of

(C) AI models and applications (C) with
(D) Imagination and thoughts (D) for

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 103

393. Two areas that show the 398. Which is the first stage in
highest growth in intelligent AI Project Cycle?
automation adoption over (A) Data Acquisition
the next three years in-
clude: (B) Project Scoping

(A) Supply chain planning and in- (C) Problem Scoping

store operations. (D) Problem Solving
(B) Store automation 399. Which of the following is
(C) Checkout considered the highest level
of intelligence?
(D) none of above
(A) Supervised machine learning
394. Which search is complete (B) Reinforced Learning
and optimal when h(n) is
consistent? (C) Unsupervised ML
(D) Reactive Machines
(A) A* search
(B) A* Tree search 400. Visual communication is
not conveyed through
(C) A* Graph search
(A) Signboards
(D) Best-first search
(B) Graphs
395. I’d like to have my shoes (C) Words
at one.
(D) Pictures
(A) repair
401. AGI stands for:
(B) repairing
(A) Artificial Global Intelligence
(C) repaired
(B) Artificial General Intelligence
(D) being repaired
(C) Artificial Ground Intelligence
396. What would be the out- (D) Artificial Globe Intelligence
put?print(”8 + 6”)
402. A natural language gener-
(A) 14 ation program must decide
(B) 8+6
(C) 86 (A) what to say
(D) (”8+6”) (B) when to say something
(C) why it is being used
397. Who is the father of Artifi-
cial Intelligence (D) both what to say & when to
say something
(A) Doug Cutting
403. Which search method will
(B) John McCarthy
expand the node that is clos-
(C) William S. est to the goal?
(D) Donald Michie (A) Best-first search

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
104 Chapter 2. Application of supervised learning

(B) A* search jeopardy because of their

(C) Greedy best-first search serious side effect.

(D) None of the mentioned (A) at risk

(B) in reality
404. DEEP LEARNING is often

NARAYAN CHANGDER
used in: (C) under control
(D) under pressure
(A) OBJECT DETECTION
(B) IMAGE PROCESSING 409. Which of the following is
not included in the recurrent
(C) TO OVERCOME LIMITA-
neural network usage sce-
TIONS OF MACHINE LEARNING
nario?
(A) Machine translation
(D) ALL OF THE ABOVE
(B) Speech recognition
405. What does AI stand for?
(C) Image style migration
(A) Artificial Intelligence
(D) Text generation
(B) Artificial Inteligence
410. systems are commonly
(C) Artifical Intelligence
used for testing theories and
(D) Artificial Intelligen learning more about the
working of human brain.
406. is defined as the per-
centage of correct predic- (A) Narrow All
tions out of all the observa- (B) Artificial General intelligence
tions.
(C) Applied Al
(A) Predictions
(D) Cognitive All
(B) Accuracy
411. Artificial intelligence sys-
(C) Reality
tems
(D) F1 Score
(A) duplicate the working of hu-
407. Select one application man brain
mentioned in the video, (B) try to copy the working of hu-
which uses NLP. man brain
(A) Twitter (C) try to partially copy the work-
(B) Facebook ing of the human brain

(C) Instagram (D) are a thing of fiction

(D) None of these 412. If according to the hypoth-

esis, the result should be
408. CLOSEST in meaningA.I. positive, but in fact it is neg-
techniques help medical doc- ative, then it is known as
tors to discover subtle in-
teractions between medica- (A) False Negative Hypothesis
tions that put patience in (B) False Positive Hypothesis

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 105

(C) Specialized Hypothesis 417. Which of the following is

a formalization of relation-
(D) Consistent Hypothesis
ships between variables in
413. A GPS navigation system the form of mathematical
for finding the fastest route. equations?
Is this AI (A) Deep Learning
(A) Yes (B) Machine Learning
(B) No (C) Statistical analysis

(C) Kind of (D) Statistical model

(D) none of above 418. OPPOSITE in meaningThe

critical issue is whether gen-
414. Whoever up with that uine artificial intelligence
idea should have won the can ever be truly realized.
Nobel Prize.
(A) authentic
(A) came (B) fake
(B) caught (C) unpretentious
(C) kept (D) true
(D) put
419. who is the first robot citi-
zen of the world
415. One downside of using AI
is that if you train it incor- (A) sophia
rectly, the AI can show pref- (B) asimo
erence for one group over
another. For example, only (C) kuri
recognizing the x-rays of (D) honda E2-DR
men and not women. This
is called 420. Set of algorithms, mod-
eled loosely after the human
(A) Biased brain, that are designed to
(B) Unbiased recognise patterns.
(C) Cheating (A) Robot calculator
(D) None of These (B) Artificial brain networks
(C) Neurons
416. Pupils should have the
(D) Neural Networks
doctor their eyes regu-
larly. 421. AI stands for
(A) tested (A) Antificial Intelligent
(B) test (B) Artificial Intelligence
(C) to test (C) Artificial Intelligent
(D) testing (D) Antificial Intelligence

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
106 Chapter 2. Application of supervised learning

422. I can do only one task, but (B) Helps us to get information in
I do it perfectly. seconds
(A) Narrow AI (C) Helps with security
(B) General AI (D) All of the above

NARAYAN CHANGDER
(C) Super AI
427. What is the other name of
(D) none of above informed search strategy?
423. The A.I have made (A) Simple search
medicine easier, more effec-
tive and efficient, an exam- (B) Online search
ple of this is (C) Heuristic search
(A) Doctors can see 3D images of (D) None of the mentioned
organs from their laptop
(B) Gives more information than 428. In May 1997, the famous
a human can detect “Human-Machine Wars” fi-
nal computer defeated Kas-
(C) Monitor their health and rou- parov, theworld chess king,
tine tests with a total score of 3.5
(D) All of them are correct to 2.5. Is this computer
called?
424. Playing chess is a task that
require (A) Dark blue
(A) Computer (B) Dark green
(B) intelligence (C) Ponder
(C) application (D) Blue sky
(D) none of above
429. The telescope will photo-
425. The following statements graph distant galaxies,
describe Artificial Intelli- attempt to understand their
gence EXCEPT past.
(A) AI is very useful in the present (A) with
generation increasing its produc-
(B) in
tivity.
(B) Al is a threat if mismanaged or (C) for
used for bad intentions. (D) on
(C) AI replaces humans as it is
more efficient than humans. 430. Which of the following is a
subset of AI?
(D) All of the above
(A) Machine learning
426. Why do we need Artificial
(B) Deep learning
Intelligence?
(C) Both a) and b)
(A) It helps with everyday prob-
lems (D) None of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 107

431. Which type of actuator 436. Which of the following is

generates a good deal of the disadvantage of artifi-
power but tends to be cial intelligence?
messy? (A) Think Logically without Bias/
(A) Electric Emotions
(B) Hydraulic (B) Machines would not be af-
(C) Thermal fected by hostile environments
(D) None (C) Robos can lead to unemploy-
ment
432. Reasoning has broadly
two types-deductive and (D) High Precision

(A) Addictive 437. is a type of book or film

(B) Logical showing space travel of life
on other planets, based on
(C) Inductive
imagined scientific discover-
(D) Calculative ies of the future.
433. Training a pet with re- (A) Epic
wards and punishments (B) Fiction
looks exactly like learn-
ing (C) Non-fiction

(A) Supervised (D) Science fiction

(B) Unsupervised 438. What is not an application
(C) Reinforcement of Artificial Intelligence?
(D) Deep (A) Self-driving cars

434. Someone is checking my (B) Recipes and cooking

car tyres for me. (C) Psychotherapy
(A) My car tyres are being (D) Voice Assistants
checked by myself.
439. Is siri considered as an
(B) I have my car tyres checked
(AI?)
regularly.
(A) True
(C) I’m having my car tyres to be
checked by someone. (B) False
(D) I’m getting my car tyres (C) Maybe
checked. (D) none of above
435. Voice to text:
440. Artificial Intelligence is a
(A) computer vision
(A) Field of computer science that
(B) robotics seeks to look at the logic of hu-
(C) Natural Language Processing man intelligence and recreate that
in machines.
(D) none of above (B) Not real

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
108 Chapter 2. Application of supervised learning

(C) Bluff 445. The first ‘intelligent’ hu-

(D) none of above manoid robot, built in Japan
in 1972 was called
441. CLOSEST in meaningAd- (A) Shakey
vanced technologies may
(B) Sophia

NARAYAN CHANGDER
make humanity insignifi-
cant both in life and work. (C) WABOT-1
(A) Future (D) Roomba
(B) Radical 446. Why is artificial intelli-
(C) Sophisticated gence different from normal
human intelligence?
(D) Unconventional
(A) As it is made to replicate hu-
442. Which of the following man intelligence
uses machine-learning tech- (B) As our technology is not as ad-
nology to become smarter vanced to replicate a human
and more capable of under-
standing natural language (C) It is made to know everything
questions and requests? making it different

(A) Netflix (D) All of the above

(B) Siri 447. What is the full form of

NLP?
(C) Cogito
(A) Natural Language Parts
(D) Deepblue
(B) Natural Learning Process
443. Which of the following (C) Natural Language Processing
types of data can be used
by artificial intelligent sys-
tems? (D) Natural Long Processing
(A) Sound data 448. What is the term used
(B) Text data to describe the intelli-
gence demonstrated by ma-
(C) Image data chines?
(D) All of these (A) artificial intelligence
444. A major thrust of AI is (B) machine learning
in the development of com- (C) natural intelligence
puter functions associated
(D) IBN
with human intelligence
(A) AI is not associated with hu- 449. Most smartphones have
man intelligence voice input ( ) function
that allows users to use
(B) False
voice typing with speech
(C) True recognition program.
(D) None of the Above (A) text-to-speech

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 109

(B) speech-to-speech 455. It is a device module, or

(C) text-to-text system used in electronic
gadgets and materials to de-
(D) speech-to-text tect events and changes.
450. NLP is: (A) Electrons
(A) Natural Language Planning (B) Motor
(B) Natural Language Preparing (C) Sensor
(C) Natural Language Processing
(D) detector

(D) None of these 456. is to read, decipher, un-

derstand, and make sense
451. Who among the follow- of the human languages in
ing is the pioneer of AI re- a manner that is valuable.
search?
(A) Natural Language Processing
(A) Google
(B) Facebook (B) Machine Learning

(C) US department of defence (C) Computer Vision

(D) US department of justice (D) none of above

452. Which is an appropriate 457. Translation:

language for describing the
(A) computer vision
relationships?
(A) First-order logic (B) robotics

(B) Propositional logic (C) Natural Language Processing

(C) ILP (inductive Logic Program-
ming) (D) none of above
(D) None of the mentioned
458. What Does VR Mean
453. Which of these is a field (A) Virtual R
that is closely related to AI?
(B) Augmented Reality
(A) Wiring
(C) Virtual Reality
(B) Drawing
(C) Mathamatics (D) VR

(D) French 459. Artificial Intelligence is

a machine specialized in one
454. Coined the term artifi-
task or area
cial intelligence.
(A) Marvin Minsky (A) General

(B) Nathaneil Rochester (B) Super

(C) Claude Shannon (C) Advance
(D) John Mc Carthy (D) Narrow

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
110 Chapter 2. Application of supervised learning

460. Why is Tesla considered as 465. What is a chat-bot?

an example of AI? (A) A robot to teach babies how to
(A) Because it is an electric pow- talk.
ered vehicle (B) An app that is used to share
(B) Because of its self driving fea- photos and videos.

NARAYAN CHANGDER
ture (C) A computer assisting software
(C) Because it is famous world used in customer services.
wide (D) none of above
(D) Because it saves the environ-
ment by not using fuel 466. A music recommendation
system such as Spotify that
461. Artificial Narrow intelli- suggests music based on the
gence is also known as: user’s listening behavior. Is
(A) Weak AI this AI
(B) Strong AI (A) Yes
(C) Moderate AI (B) No
(D) None of these (C) Kind of
(D) none of above
462. When is the term Artificial
Intelligence applied? 467. SDGs are the collection of
(A) When the machines mimic the interlinked goals.
cognitive functions that humans (A) 16
ought to carryout (B) 17
(B) When a machine turns on (C) 18
(C) When a machine performs (D) 19
very well
(D) When a machine becomes 468. By Problem Scoping, we
faulty mean
(A) understanding the problem
463. Who coined the term “Ar-
tificial Intelligence”? (B) examining who will be af-
fected
(A) John McCarthy
(C) checking where can this help
(B) Hebert Simon
(D) all of the above
(C) Allen Newell
(D) Allan Turing 469. Unluckily, David’s danger-
ous and incredible journey
464. Which of the following is in search of the Blue Fairy is
not the type of AI in
(A) Reactive machines (A) advance
(B) Unlimited memory (B) danger
(C) Self-awareness (C) end
(D) Theory of mind (D) vain

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.12 Artificial Intelligence 111

470. Humans use this to com- 475. The role of neural net-
municate with the computer works comes under which
and tell it to do something. type of learning?
(A) Programming Language (A) Supervised
(B) Program (B) Unsupervised
(C) Output (C) Reinforcement

(D) Command (D) Deep Learning

476. Which of the following is
471. Weak artificial intelligent
not a module in the Tensor-
systems
Flow library?
(A) show intelligent behaviour (A) tf.nn
(B) lack natural language pro- (B) tf. layers
cessing skills
(C) tf.contrib
(C) cannot perform complex tasks
(D) tf. boost

(D) can achieve self-consciousness 477. Is Doraemon an Artificial

with time Intelligence?
(A) YES
472. Pupils should have their
(B) NO
eyes regularly.
(C) I don’t no
(A) tested
(D) none of above
(B) to test
478. With the help of com-
(C) test
puters understand the com-
(D) testing mon language used by hu-
man beings.
473. What do chatbots use?
(A) Neural Language Processing
(A) AI & natural language pro-
(B) Common Language Process-
cessing
ing
(B) Cartoon characters (C) Natural Language Processing
(C) Talkative people
(D) none of above (D) Human Language Processing

474. Which of the following sci- 479. WHAT IS THE FULL FORM
entist who were involved in OF AGI?
the project of AI? (A) ARTIFICIAL GENERAL INTELLI-
(A) Albert Einstein GENCE
(B) ARTIFICIAL GRAND INTELLI-
(B) Marvin Minsky
GENCE
(C) Charles Darwin
(C) ARTIFICIAL GRASP INTELLI-
(D) Gregor Mendel GENCE

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
112 Chapter 2. Application of supervised learning

(D) none of above 485. The best AI agent is one

which
480. Which of the following
artificial intelligence sys- (A) Needs user inputs for solving
tems are not limited to pro- any problem
grammed responses?

NARAYAN CHANGDER
(B) Can solve a problem on its
(A) Neural Network Systems own without any human interven-
tion
(B) Narrow Al Systems
(C) Strong Al Systems (C) Need a similar exemplary
problem in its knowledge base
(D) Special Al Systems
(D) All of the above
481. Which of the following are
the terms used under artifi- 486. A type of AI which is able
cial intelligence? to find a solution without hu-
(A) Robotics man intervention is:

(B) Machine Learning (A) Weak Artificial Intelligence

(C) Big Data (B) Strong Artificial Intelligence
(D) All of the above (C) Computer Language

482. is a set of technolo- (D) Machine Language

gies that allows us to extract
487. Which of these does not
knowledge from data.
use NLP?
(A) Extract knowledge
(A) Translation Tools
(B) Artificial intelligence
(B) Email Filters
(C) Direct intelligence
(C) Facial Recognition
(D) Machine learning
(D) Autocomplete Feature
483. Web Crawler is an exam-
ple of 488. External actions of the
(A) Intelligent Agent agent is selected by
(B) Problem-solving agent (A) Performance
(C) Simple reflex agent (B) Perceive
(D) Model-based agent (C) Learning
484. Which of the following (D) Actuator
is not one of the three
domains of artificial intelli- 489. Voice assistants are which
gence? form of AI?

(A) Data (A) Strong AI

(B) Natural Language Processing (B) Narrow AI
(C) Neural Networks (C) Week AI
(D) Computer vision (D) General AI

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 2.13 Machine learning(HARD QUESTIONS 113

490. Which of the following (B) Sensors

means ‘Cleaning of collected
data for the project’? (C) Data Wrangling
(A) Data Collection (D) Data Analysis

2.13 Machine learning(HARD QUESTIONS

1. Among the following option 5. How will you check data
identify the one which is types in X train?
used to create the most com- (A) x train.dtypes
mon graph types.
(B) X train.datatypes
(A) qplot
(C) X train.dtypes
(B) plot
(D) none of these
(C) quickplot
(D) All of the above qplot is used 6. Suppose we would like to
to create the most common graph convert a nominal attribute
types X with 4 values to a data
table with only binary vari-
2. In dataset loading, The vari- ables. How many new at-
ables of data are called its? tributes are needed?
(A) Response (A) 4
(B) Features (B) 8
(C) Target (C) 1
(D) Vector
(D) 12
3. What is the purpose of this
7. In Random Forest the Mem-
code?print ( ‘Accuracy:%.3f’
ory requirement for the stor-
% accuracy score(x
age process?
pred, y test))
(A) print the accuracy (A) High Memory

(B) finding the accuracy (B) Low Memory

(C) accuracy score (C) No Memory

(D) none of this (D) None of theAbove

4. What is the use of this code? 8. Application of machine

print (digits.DESCR) learning methods to large
databases is called
(A) This will print the description
of digits. (A) data mining.
(B) print numbers (B) artificial intelligence
(C) none of this (C) big data computing
(D) none of above (D) internet of things

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
114 Chapter 2. Application of supervised learning

9. How many types of Machine (B) show the lable of x as height

Learning Techniques? and y as count
(A) 3 (C) x and y will be height and
(B) 5 count

NARAYAN CHANGDER
(D) none of the above
(C) 7
(D) 9 There are three(3) types 14. Example of Reinforcement
of Machine Learning techniques, learning
which are Supervised Learning, (A) chess game
Unsupervised Learning, and Rein-
forcement Learning . (B) object recognition
(C) Weather conditions
10. Which library is to be
imported for creating (D) price of house
DataFrame? 15. What is the role of Ex-
(A) Python ploratory Data Analysis
(B) DataFrame (EDA) in data analysis?

(C) Pandas (A) Making sense of database

(D) Random (B) To discover patterns

(C) To spot anomalies
11. Which of the following pa-
(D) All of these
rameter is used to specify
row or column in rename 16. what is the use of this code?
function of DataFrame? tree.plot tree(clf)
(A) rowindex (A) We can visualize the tree using
(B) colindex tree.plot tree
(C) Both of the above (B) We can classify the tree using
tree.plot tree
(D) index
(C) We can plot the tree using
12. is basically a type tree.plot tree
of unsupervised learning
(D) all of these
method
(A) Unsupervised learning 17. In Decision Tree algorithms
entropy of a given dataset is
(B) clustering
zero. This statement implies
(C) semi supervised
(D) classification clustering is basi- (A) Further splitting is required
cally a type of unsupervised learn-
(B) Need some other information
ing method.
to decide splitting
13. What is use of this code? (C) No further splitting is required
‘Height’ ) ‘Count’ )
(A) x as height and y as count (D) None of the Mentioned

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.13 Machine learning(HARD QUESTIONS 115

18. hich one is the correct syn- 23. Imagine a Newly-Born

tax to find the unique values starts to learn walking. It
in the Year column from the will try to find a suitable
car dataset? policy to learn walking af-
(A) car[’year’].unique==True ter repeated falling and get-
ting up.specify what type
(B) car [’year’]=unique() of machine learning is best
(C) car [’year’].unique() suited?
(D) car [’year’]=car.year.unique() (A) classification
19. refers to converting the (B) regression
labels into numeric form so (C) kmeans algorithm
as to convert it into the
machine-readable form. (D) reinforcement learning
(A) Euclidean Distance
24. in Read CSV Files what is
(B) Min-Max Normalization meant by CSV?
(C) Minkowski Distance (A) comma separated files
(D) Label Encoding
(B) copied separated files
20. NumPY stands for? (C) comma simple files
(A) Numbering Python (D) none of these
(B) Number In Python
(C) Numerical Python 25. We can create DataFrame
from
(D) None Of the above
(A) Numpy arrays
21. In NumPy, what does the
(B) List of Dictionaries
SHAPE of an array mean?
(A) The shape is the number of (C) Dictionary of Lists
columns. (D) All of the above
(B) The shape is the number of
rows 26. How will you show the
relationship between the
(C) The shape is the number of el-
columns?
ements in each dimension.
(D) all of these (A) import pandas as pddf =
pd.read csv(’data.csv’)print(df.corr())
22. How can we find the num-
ber of rows and columns in (B) import pandas as pdprint(df.correlation())
the data?
(A) dataframe.shape()
(C) import pandas as pddf =
(B) dataframe.size() pd.read csv(’data.csv’)print(df.corre())
(C) dataframe.shape
(D) dataframe.size (D) all of these

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
116 Chapter 2. Application of supervised learning

27. Machine Learning is a sub- 32. Which of the following will

set of be true about k in K-Nearest
(A) Data Learning Neighbor in terms of Bias?

(B) Deep Learning (A) When you decrease the k the

bias will be increases

NARAYAN CHANGDER
(C) Artificial Intelligence
(B) When you increase the k the
(D) all of these bias will be increases
28. In Decision Tree, is divid- (C) Both (A) and (B)
ing the root node into dif-
(D) None of the Above When you
ferent parts on the basis of
increase the k the bias will be in-
some condition.
creases
(A) pruning
(B) splitting 33. is a metric to measure
how often a randomly cho-
(C) leaf node sen element would be incor-
(D) parent node rectly identified.
(A) Information Gain
29. How to check missing val-
ues in categorical variables (B) Entropy
(A) df[categorical].isnull(). sum () (C) none of these
(B) df[categorical].isnull() (D) Gini Index Gini Index is a met-
(C) df[categorical]. sum () ric to measure how often a ran-
domly chosen element would be
(D) df[categorical].null(). sum () incorrectly identified
30. Amongst which of the fol-
34. Full form of PAC is
lowing can be used to create
various inputs using pandas (A) Probably Approximate Cost
DataFrame. (B) Probably Approximate Cor-
(A) Lists, dict rect.
(B) Series (C) Probably Approximate Com-
(C) Numpy ndarrays and Another munication
DataFrame (D) Probably Approximate Com-
(D) All of the above mentioned putation PAC stands for Probably
Approximate Correct.
31. The distance between two
points calculated using 35. Spam Classification is an ex-
Pythagoras theorem is ample for?
(A) supremum distance (A) Naive Bayes
(B) eucledian distance (B) Probabilistic condition
(C) linear distance (C) Random Forest
(D) manhattan distance (D) All the Above

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
2.13 Machine learning(HARD QUESTIONS 117

36. What is a correct syntax to 40. We can create a NumPy

return the first value of a object by using the array()
Pandas Series? function.
(A) pd.myseries[mylist] (A) ndarray
(B) pd.createmyseries[mylist] (B) nparray
(C) numpyarray
(C) pd.getmyseries[mylist]
(D) none of these
(D) none of the above
41. What does the GraphViz
37. What does this code gives?
do? car[’name’] =
(A) GraphViz gives a better and
car[’name’].str.split(’ ‘).str.slice(0,
clearer Graph.
3).str.join(’ ‘)
(B) GraphViz gives detailed graph
(A) Splitting the name into a list
and removing the first two words
from the list (C) GraphViz gives only given val-
ues graph
(B) Getting the first two words
from the name (D) none of these

(C) Splitting the name into the list 42. Which type of learning algo-
rithm used for “future crude
oil prices ‘’?
(D) None of these
(A) Recognizing similarities
38. Identify the kind of learning (B) predicting
algorithm for “facial identi- (C) creatind patterns
ties for facial expressions”.
(D) all the above
(A) Prediction
43. During the execution of
(B) Recognition Patterns
following code, what will
(C) Recognizing anomalies be the response, we get
import pandas as pds
(D) Generating Patterns
=pd.Series([1, 2, 3, 4, 5],
index= [’a’, ‘b’, ‘c’, ‘d’,
39. for var in categori-
‘e’])print(s[’f’])
cal:print(df[v0ar].value
counts())What does this (A) KeyError
code means? (B) IndexError
(A) To view the values (C) ValueError
(B) view frequency of categorical (D) None of the above mentioned
variables
44. Machine Learning is a field
(C) view all the values and vari- of AI consisting of learning
ables algorithms that
(D) All of these (A) At executing some task

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
118 Chapter 2. Application of supervised learning

(B) Over time with experience (C) Classification

(C) Improve their performance (D) none of these
(D) All of the above 49. dec= round(decision
45. The way to install the pan- tree.score(x train, y

NARAYAN CHANGDER
das library train) * 100, 2)
(A) Return the sample value
(A) install pandas
(B) Return the observations
(B) pandas install python
(C) Return the Accuracy
(C) python install pandas
(D) None of these
(D) None of the above
50. How to check number of
46. Automatic recognition of
unique values from each col-
patterns in data (text, hand-
umn pandas
writing, images, sound and
video) (A) nunique()
(A) Recognition (B) . unique()
(B) Training (C) .nunique()
(C) Predictive models (D) both 1 and 3
(D) Testing 51. The Scikit Learn library uses
the method to infer the
47. Clustering is and is exam-
properties of data
ple of learning
(A) extract()
(A) predictive and supervised
(B) encode()
(B) predictive and unsupervised
(C) transform()
(C) descriptive and supervised
(D) classify()
(D) descriptive and unsupervised
52. RGB image is represented
48. A Machine Learning tech- by size matrix.
nique that helps in detecting
the outliers in data.what is (A) row X column
it called as?? (B) row X column X 2
(A) Clustering (C) row X column X k
(B) Anomaly Detection (D) row X column X 3

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
 Index

A Bias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Binary attribute . . . . . . . . . . . . . . . . . . 16
Abilities . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Binary classification . . . . . . . . . . . . . . 42
Ability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Binary variables . . . . . . . . . . . . . . . . 113
Actions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Biological neural networks . . . . . . . 48
Activation . . . . . . . . . . . . . . . . . . . . . . . . 37 Business requirements . . . . . . . . . . . 61
Activation function . . . . . . . . . . . . . . . 36
Actual positive values . . . . . . . . . . . . 59
Adult human intelligence . . . . . . . . . 82 C
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . 3
Algorithm learns . . . . . . . . . . . . . . . . . 35 Cars. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 9 Categorical variables . . . . . . . . . . . 117
Algorithms work . . . . . . . . . . . . . . . . . 45 Categories. . . . . . . . . . . . . . . . . . . . . . . 25
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Cell body . . . . . . . . . . . . . . . . . . . . . . . . 45
Artificial intelligence timeline . . . . . 81 Certain actions . . . . . . . . . . . . . . . . . . 19
Artificial intelligences. . . . . . . . . . . . . 65 Champions . . . . . . . . . . . . . . . . . . . . . . 98
Artificial narrow intelligence . . . . . . 67 Change . . . . . . . . . . . . . . . . . . . . . . . . . 17
Artificial neural network . . . . . . . . . . 36 Chess . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Artificial neural networks . . . . . . . . . 46 Class neural network . . . . . . . . . . . . 25
Classes . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Classification model . . . . . . . . . . . . . 20
B Classification type . . . . . . . . . . . . . . . 26
Cluster . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Backward links . . . . . . . . . . . . . . . . . . 45 Clustering. . . . . . . . . . . . . . . . . . . . . . . . . 9
Batch normalization . . . . . . . . . . . . . 53 Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Behavior. . . . . . . . . . . . . . . . . . . . . . . . . 89 Common graph types . . . . . . . . . . 113
Best choice . . . . . . . . . . . . . . . . . . . . . . 11 Complex problems . . . . . . . . . . . . . . 39
Best robot . . . . . . . . . . . . . . . . . . . . . . 101 Computational intelligence. . . . . . . 90
Better performance . . . . . . . . . . . . . . 53 Computational model. . . . . . . . . . . . 44

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
120 INDEX

Computers . . . . . . . . . . . . . . . . . . . . . . . . 4 Error rate . . . . . . . . . . . . . . . . . . . . . . . . 29

Confusion matrix . . . . . . . . . . . . . . . . 20 Event . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Continuous dependent variable . . . 3 Events . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Continuous output variable . . . . . . 32 Example . . . . . . . . . . . . . . . . . . . . . . . . . 11
Continuous variables . . . . . . . . . . . . 31 Experience . . . . . . . . . . . . . . . . . . . . . . . 32

NARAYAN CHANGDER
Convolution output . . . . . . . . . . . . . . 51 Exponential technological
Convolutional layer . . . . . . . . . . . . . . 45 development . . . . . . . . . . . . . 64
Correct option . . . . . . . . . . . . . . . . . . . 88 Eyes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Correct order . . . . . . . . . . . . . . . . . . . . 45
Correction . . . . . . . . . . . . . . . . . . . . . . . 26
Corrupted data . . . . . . . . . . . . . . . . . . 26 F
Customers . . . . . . . . . . . . . . . . . . . . . . . 12
Faces . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Facial expressions . . . . . . . . . . . . . . . 20
D Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
False negatives. . . . . . . . . . . . . . . . . . . . 7
Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 False positives . . . . . . . . . . . . . . . . . . . 12
Databases . . . . . . . . . . . . . . . . . . . . . . . 78 Fastest route. . . . . . . . . . . . . . . . . . . . 105
Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Feature map . . . . . . . . . . . . . . . . . . . . . 49
Dataset loading . . . . . . . . . . . . . . . . 113 Final prediction . . . . . . . . . . . . . . . . . . 23
Day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Fish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Decades . . . . . . . . . . . . . . . . . . . . . . . . . 65 Following questions . . . . . . . . . . . . . . 67
Decisions . . . . . . . . . . . . . . . . . . . . . . . . 38 Form. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Deep learning library . . . . . . . . . . . . 60 Fully connected layer. . . . . . . . . . . . . 55
Deep learning models . . . . . . . . . . . 52 Functions . . . . . . . . . . . . . . . . . . . . . . . . 42
Deep neural network . . . . . . . . . . . . 54 Future . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Deeper data . . . . . . . . . . . . . . . . . . . . . 83 Futuristic actions . . . . . . . . . . . . . . . . . 41
Dependent variable . . . . . . . . . . . . . . . 6
Detailed steps . . . . . . . . . . . . . . . . . . . 91
Deterministic algorithm . . . . . . . . . . 28
G
Development . . . . . . . . . . . . . . . . . . . . 60 General learning . . . . . . . . . . . . . . . . 58
Different art styles. . . . . . . . . . . . . . . . 92 General-purpose learning algorithms
Different human languages . . . . . . 71 51
Different sound . . . . . . . . . . . . . . . . . . 70 Generalization error . . . . . . . . . . . . . 29
Discrete categorical outputs . . . . . . . 9 Genuine artificial intelligence . . . 105
Distant galaxies . . . . . . . . . . . . . . . . 106 Given dataset. . . . . . . . . . . . . . . . . . . . 16
Distribution value . . . . . . . . . . . . . . . . 27 Given options. . . . . . . . . . . . . . . . . . . . 50
Domains . . . . . . . . . . . . . . . . . . . . . . . . 67 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Good test dataset . . . . . . . . . . . . . . . . 29
Grade science test . . . . . . . . . . . . . . . 84
E Gradient . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Gradients. . . . . . . . . . . . . . . . . . . . . . . . 50
Electronic person . . . . . . . . . . . . . . . . 80
Ensemble method. . . . . . . . . . . . . . . . 55
Environment . . . . . . . . . . . . . . . . . . . . . 20 H
Environmental pollution . . . . . . . . . . 63
Equation. . . . . . . . . . . . . . . . . . . . . . . . . 33 Hard margin . . . . . . . . . . . . . . . . . . . . 29

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
INDEX 121

Heads . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Help. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Layers exploits . . . . . . . . . . . . . . . . . . . 57
Hidden layer. . . . . . . . . . . . . . . . . . . . . 35 Learning . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Human . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Learning algorithm . . . . . . . . . . . . . . . . 4
Human brain . . . . . . . . . . . . . . . . . . . . 42 Learning model . . . . . . . . . . . . . . . . . . 23
Human capabilities . . . . . . . . . . . . . 100 Limited functional area. . . . . . . . . . . 65
Human emotions . . . . . . . . . . . . . . . . 65 Linear . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Human intelligence . . . . . . . . . . . . . . 62 Linear constant value . . . . . . . . . . . . 41
Human intervention . . . . . . . . . . . . 112 Linear relationship . . . . . . . . . . . . . . . 28
Human language . . . . . . . . . . . . . . . . 94 Logistic regression analysis . . . . . . . 33
Human speech . . . . . . . . . . . . . . . . . . 93 Long-term sequence data . . . . . . . . 60
Human-level performance . . . . . . . 45 Loss functions . . . . . . . . . . . . . . . . . . . . 53
Hypothesis. . . . . . . . . . . . . . . . . . . . . . . 16 Low dimensional data . . . . . . . . . . . 30

I M
Idea . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Machine . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Image classification problem . . . . . 66 Machine learning . . . . . . . . . . . . . . . . . 4
Image recognition problem . . . . . . 58 Machine learning algorithms . . . . . . 3
Images . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Machine learning model . . . . . . . . . 17
Import pandas . . . . . . . . . . . . . . . . . 115 Machine translation . . . . . . . . . . . . . . 55
Independent variables . . . . . . . . . . . 13 Machine-learning technology . . . 108
Individual base models . . . . . . . . . . 61 Machine-readable form . . . . . . . . 115
Information . . . . . . . . . . . . . . . . . . . . . . 10 Major breakthrough . . . . . . . . . . . . . 80
Input data . . . . . . . . . . . . . . . . . . . . . . . . 6 Mathematical equations . . . . . . . . 105
Input layer . . . . . . . . . . . . . . . . . . . . . . . 41 Max inputs. . . . . . . . . . . . . . . . . . . . . . . 60
Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Max pooling . . . . . . . . . . . . . . . . . . . . . 59
Integrated learning strategies . . . . 88 Memory . . . . . . . . . . . . . . . . . . . . . . . . . 39
Intelligence . . . . . . . . . . . . . . . . . . . . . . . 4 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Iteration . . . . . . . . . . . . . . . . . . . . . . . . . 25 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Minimum distance . . . . . . . . . . . . . . . 17
Minute. . . . . . . . . . . . . . . . . . . . . . . . . . . 63
K Missing values. . . . . . . . . . . . . . . . . . 116
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Key points . . . . . . . . . . . . . . . . . . . . . . . 41 Model parameter . . . . . . . . . . . . . . . . 17
Knowledge . . . . . . . . . . . . . . . . . . . . . . 59 Models . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Multiple regression. . . . . . . . . . . . . . . 32
Multiple regression model. . . . . . . . 30
L
Labelled data . . . . . . . . . . . . . . . . . . . . 12 N
Language . . . . . . . . . . . . . . . . . . . . . . . . . 6
Languages . . . . . . . . . . . . . . . . . . . . . . 75 Name . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Large databases. . . . . . . . . . . . . . . . 113 Network . . . . . . . . . . . . . . . . . . . . . . . . . 10
Large penalty . . . . . . . . . . . . . . . . . . . . 32 Network architecture . . . . . . . . . . . . . 38
Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Neural network . . . . . . . . . . . . . . . . . . 10

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
122 INDEX

Neural networks . . . . . . . . . . . . . . . . . 38 Primitive forms . . . . . . . . . . . . . . . . . . . 66

Neuron . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Probabilities . . . . . . . . . . . . . . . . . . . . . 59
New planet . . . . . . . . . . . . . . . . . . . . . . 96 Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Nodes . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Process human languages . . . . . . . 65
Non-linear data . . . . . . . . . . . . . . . . . . . 5 Product recommendations . . . . . . . 79

NARAYAN CHANGDER
Non-vegetarian dishes . . . . . . . . . . . 75 Program . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Null hypothesis . . . . . . . . . . . . . . . . . . 13 Programming . . . . . . . . . . . . . . . . . . . . 42
Number . . . . . . . . . . . . . . . . . . . . . . . . . 10 Propagation . . . . . . . . . . . . . . . . . . . . . 52
Numerous data . . . . . . . . . . . . . . . . . . 63

Q
O
Questions . . . . . . . . . . . . . . . . . . . . . . . 71
Objective function . . . . . . . . . . . . . . . 51
Observation . . . . . . . . . . . . . . . . . . . . . 14
Option . . . . . . . . . . . . . . . . . . . . . . . . . . 27 R
Options i . . . . . . . . . . . . . . . . . . . . . . . . 23
Order . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Outcome . . . . . . . . . . . . . . . . . . . . . . . . 31 Recognise patterns . . . . . . . . . . . . . 105
Outliers . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Recognition . . . . . . . . . . . . . . . . . . . . . . 15
Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Regression equation. . . . . . . . . . . . . . . 8
Output layer . . . . . . . . . . . . . . . . . . . . . 35 Regression line . . . . . . . . . . . . . . . . . . 33
Output layers . . . . . . . . . . . . . . . . . . . . 44 Regular browsing habits . . . . . . . . . 91
Output related data . . . . . . . . . . . . . . 42 Reinforcement learning . . . . . . . . . . . . 5
Output values . . . . . . . . . . . . . . . . . . . 29 Relationship. . . . . . . . . . . . . . . . . . . . . . . 8
Relationships . . . . . . . . . . . . . . . . . . . . 67
Relevant details . . . . . . . . . . . . . . . . . . 68
P Representation learning algorithm 52
Response . . . . . . . . . . . . . . . . . . . . . . . . 51
Padding . . . . . . . . . . . . . . . . . . . . . . . . . 38 Response systems . . . . . . . . . . . . . . . . 63
Parameters . . . . . . . . . . . . . . . . . . . . . . 10 Rewards . . . . . . . . . . . . . . . . . . . . . . . . . 70
Parsimonious model . . . . . . . . . . . . . 34 Robot. . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Patient information. . . . . . . . . . . . . . . 83 Robots . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Patterns. . . . . . . . . . . . . . . . . . . . . . . . . . 20 Root node . . . . . . . . . . . . . . . . . . . . . . 116
People . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Perceptron . . . . . . . . . . . . . . . . . . . . . . . 26
Perceptron neural networks . . . . . . 44 S
Performance . . . . . . . . . . . . . . . . . . . . . . 8
Personal digital assistant . . . . . . . . . 75 Science . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Pixels . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Scientist . . . . . . . . . . . . . . . . . . . . . . . . . 62
Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Search algorithm . . . . . . . . . . . . . . . . 71
Possible paths . . . . . . . . . . . . . . . . . . . 31 Search strategy . . . . . . . . . . . . . . . . . 106
Pre-defined functions . . . . . . . . . . . . 86 Second . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Prediction output . . . . . . . . . . . . . . . . . 49 Sense. . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Predictions . . . . . . . . . . . . . . . . . . . . . . . 33 Setup code . . . . . . . . . . . . . . . . . . . . . . 37
Previous computations . . . . . . . . . . . 41 Simple approach . . . . . . . . . . . . . . . . 11
Previous layer . . . . . . . . . . . . . . . . . . . . 45 Single batch . . . . . . . . . . . . . . . . . . . . . 43

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/
INDEX 123

Single cluster . . . . . . . . . . . . . . . . . . . . 14 U
Size matrix . . . . . . . . . . . . . . . . . . . . . 118
Smart phone. . . . . . . . . . . . . . . . . . . . . 71 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Social networking sites . . . . . . . . . . . 69 Unlabeled data . . . . . . . . . . . . . . . . . . 13
Solve problem . . . . . . . . . . . . . . . . . . . 72 Unsupervised learning . . . . . . . . . . . . 4
Space . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Unsupervised learning method. . 114
Specific depth. . . . . . . . . . . . . . . . . . . . 46 Unsupervised learning type . . . . . . 24
Speech recognition . . . . . . . . . . . . . . . . 9 Unsupervised task . . . . . . . . . . . . . . . 16
Statement . . . . . . . . . . . . . . . . . . . . . . . 14 Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Statements. . . . . . . . . . . . . . . . . . . . . . . 15 Useful patterns . . . . . . . . . . . . . . . . . . 72
States. . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 User . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Stochastic gradient descent . . . . . . 55 Users . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Strong intelligence . . . . . . . . . . . . . . . 84
Systems. . . . . . . . . . . . . . . . . . . . . . . . . . 39 V
Validation . . . . . . . . . . . . . . . . . . . . . . . 11
Variables . . . . . . . . . . . . . . . . . . . . . . . . 12
T Variance . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Vector machines . . . . . . . . . . . . . . . . . 27
Target. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Video . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Task . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Virtual reality . . . . . . . . . . . . . . . . . . . . 67
Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Voice recognition . . . . . . . . . . . . . . . . 40
Technique . . . . . . . . . . . . . . . . . . . . . . . 10 Volume . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Techniques . . . . . . . . . . . . . . . . . . . . . 104
Technology . . . . . . . . . . . . . . . . . . . . . . 62
Temperature . . . . . . . . . . . . . . . . . . . . 102 W
Text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Text analysis . . . . . . . . . . . . . . . . . . . . . 58 Washing machine . . . . . . . . . . . . . . . . 74
Texts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Theworld chess king . . . . . . . . . . . . 106 Weight. . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Weights. . . . . . . . . . . . . . . . . . . . . . . . . . 42
Time series . . . . . . . . . . . . . . . . . . . . . . 59 Word . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Times. . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Words . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
World . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Tiny robots . . . . . . . . . . . . . . . . . . . . . . 89
Today. . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Training . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Y
Training process. . . . . . . . . . . . . . . . . . . 6
Trajectory . . . . . . . . . . . . . . . . . . . . . . . . 72 Y train . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Y value . . . . . . . . . . . . . . . . . . . . . . . . . . 28
True y value. . . . . . . . . . . . . . . . . . . . . . 61 Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

ALL ANSWERS UPLOADED ON OUR OFFICIAL WEBSITE

https://www.gatecseit.in/