M2Lab
Machine Learning
(112-1: EE5184)
劉子毓 Joyce Liu
1
M2Lab
Outline
● Logistics
○ Course design and rubric
○ Can I add the Machine Learning course?
● Self-introduction among students
● What is AI? What is Machine Learning?
● Major categories in Machine Learning
2
M2Lab
Outline
● Logistics
○ Course design and rubric
○ Can I add the Machine Learning course?
● Self-introduction among students
● What is AI? What is Machine Learning?
● Major categories in Machine Learning
3
M2Lab
Objective
● Explain the mathematics behind the Machine Learning models.
● Implement Machine Learning methods and apply them to real applications
● Analyze and critique the numerical results
4
M2Lab
Course Design
● Language
○ This course will be offered in English, including the homeworks and exams.
○ Students can ask questions in the language they feel most comfortable with.
● Class arrangement
○ This is a 4-credit course, taking place every Wednesday 9:10 am - 13:10 pm.
○ The first 3 hours (9:10 am - 12:10 pm) will be the lectures.
○ The last hour (12:20-13:10) will be the TA session.
■ Explain the new assignments and discuss last week’s homework.
■ The assignments include take-home quizzes implementation on colab.
■ You are encouraged to bring your laptop to the TA session to test out the Colab notebook
template and get support from the TA or other classmates.
● Grading: [30%] midterm, [30%] final exam, [40%] homework
● Office hour: Friday 9 am - 10 am @ EE2-342
5
M2Lab
Course Design
Week # Date Topic Homework
1 2023/9/6 Introduction and rubric review; What is AI? What is ML? No homework
2 2023/9/13 Regression; Cross validation; Hyperparameter optimization HW 1: Colab experiment: Find the top Youtubers
3 2023/9/20 Support vector machine (SVM); the kernel trick HW 2: Take-home quizzes
4 2023/9/27 Decision tree, Naive Bayes classifier HW 3: Colab experiment: Breast cancer classification
5 2023/10/4 Ensemble methods, random forest and AdaBoost HW 4: Take-home quizzes
6 2023/10/11 Dimension reduction: PCA, MDS, tSNE HW 5: Colab experiment: Single cell data visualization
7 2023/10/18 GMM as an example of Bayesian modeling No homework
8 2023/10/25 Midterm No homework
9 2023/11/1 Multi layer perceptron (MLP) HW 6: Take-home quizzes
10 2023/11/8 Convolutional neural network (CNN) HW 7: Colab experiment: MNIST classification
11 2023/11/15 Autoencoder HW 8: Take-home quizzes
12 2023/11/22 Variational autoencoder (VAE) HW 9: Colab experiment: MNIST image reconstruction
13 2023/11/29 Generative adversarial network (GAN) HW 10: Take-home quizzes
14 2023/12/6 Transformer HW 11: Colab experiment: Pet's Facial Expression
15 2023/12/13 Guest lecture No homework
16 2023/12/20 Final exam No homework 6
M2Lab
Homework
● The homework will be posted on the NTU COOL platform. TA will explain the
details during the 4th hours each week. You have one week to finish the
assignments.
● Take-home quizzes
○ Finish quizzes and submit your answers on NTU COOL before the deadline.
● Colab experiments
○ Set up the experiment using Colab notebooks.
○ Once finished, download the notebook (.ipynb file) and upload it to NTU COOL before the
deadline.
7
M2Lab
Colab experiments
● Log into your google account. If you don’t have one yet, read the instructions
provided by Computer and Information Networking Center.
● Create a Colab notebook for the assignment. Once you finish the experiment,
upload the .ipynb file to the NTU COOL platform.
● More detailed instructions will be provided by our TA next week.
8
M2Lab
Colab experiments
● Rubric
Criteria Ratings
Code readability 0 pts: No markdowns 20 pts: Needs improvement, e.g., some 25 pts: Excellent
comments. markdowns or comments missing.
Description of the 0 pts: No description of the 20 pts: Needs improvement, e.g., certain 25 pts: Excellent
methodology methodology. steps of the methods not described.
Accuracy 0 pts: Results are not making 20 pts: Needs improvement, e.g., failing 25 pts: Excellent
sense. some minor corner cases.
Conclusions and 0 pts: No conclusions of 20 pts: Needs improvement, listing pros and 25 pts: Excellent
discussions discussions provided. cons, but no proposal on next steps.
Total points: 100
● Let’s take a look at an example now.
9
M2Lab
Guest lecture
● Each semester, we will invite an external speaker
to share his or her experiences of applying
Machine Learning to real world challenges.
● This year, we are inviting Fangyao (Jeffrey) Hu
to share his experiences in Digital Pathology.
● Dr. Hu received his BS from NTUEE, his PhD
from Duke University. He is currently a senior
principal AI scientist at Genentech.
10
M2Lab
Can I add the Machine Learning Course?
● If you were not able to register and would like to add
the course, follow these steps.
● Depending on the total number of registered
students, we may be able to take a few more. Not
guaranteed though.
● Fill out the google form
○ Google form link: https://forms.gle/KMqgQSNhyLU5QicG8
○ QR code is on the right.
● If your request is granted, we will email you the
registration code.
11
M2Lab
Can I add the Machine Learning Course?
● If you were not able to register and would like to add
the course, follow these steps.
● Depending on the total number of registered
students, we may be able to take a few more. Not
guaranteed though.
● Fill out the google form
○ Google form link: https://forms.gle/HTa2jC34cQ3JXkwf6
○
○ QR code is on the right.
● If your request is granted, we will email you the
registration code.
12
M2Lab
Outline
● Logistics
○ Course design and rubric
○ Can I add the Machine Learning course?
● Self-introduction among students
● What is AI? What is Machine Learning?
● Major categories in Machine Learning
13
M2Lab
Self-introduction among students
During the course of studies, you likely will exchange thoughts on the homeworks
or engage in small group discussion with other students. We have a large group of
students with diverse backgrounds. To break the ice, please tell us about yourself.
● Your name.
● Why did you choose to come to this class?
● What do you expect to get out of this class?
● A little known fact about yourself.
14
M2Lab
Outline
● Logistics
○ Course design and rubric
○ Can I add the Machine Learning course?
● Self-introduction among students
● What is AI? What is Machine Learning?
● Major categories in Machine Learning
15
M2Lab
Artificial Intelligence (AI) & Machine Learning (ML)
Discuss with your neighboring classmates about the following questions.
[10 minutes]
● What is AI?
● What is Machine Learning (ML)?
16
M2Lab
Artificial Intelligence (AI) & Machine Learning (ML)
“Artificial intelligence (AI) is concerned with building systems that
simulate intelligent behavior. It encompasses a wide range of approaches,
including those based on logic, search, and probabilistic reasoning.
Machine learning (ML) is a subset of AI that learns to make decisions by
fitting mathematical models to observed data. This area has seen explosive
growth and is now (incorrectly) almost synonymous with the term AI. A
deep neural network is one type of machine learning model, and when this
model is fitted to data, this is referred to as deep learning.”
Prince, Simon JD. UNDERSTANDING DEEP LEARNING. MIT PRESS, 2023.
17
M2Lab
Artificial Intelligence (AI) & Machine Learning (ML)
A system that simulate intelligent
behavior, relying on logic, search,
AI and probabilistic reasoning.
ML
Make decisions by
fitting mathematical DL
models to observed A type of ML model with
data. deep neural network
architecture fitted to data.
18
M2Lab
Artificial Intelligence (AI) & Machine Learning (ML)
A system that simulate intelligent
behavior, relying on logic, search,
AI and probabilistic reasoning.
ML The focus of this course.
Make decisions by
fitting mathematical DL
models to observed A type of ML model with
data. deep neural network
architecture fitted to data.
19
M2Lab
Outline
● Logistics
○ Course design and rubric
○ Can I add the Machine Learning course?
● Self-introduction among students
● What is AI? What is Machine Learning?
● Major categories in Machine Learning
20
M2Lab
Major categories in ML
There are three major categories.
● Supervised learning: “Building a statistical model for predicting or estimating
an output based on one or more imputes.”
○ James, Gareth, et al. An introduction to statistical learning. Vol. 112. New York: springer, 2013.
● Unsupervised learning: “A statistical model built to learn the relationships and
structure without the supervising output.”
○ James, Gareth, et al. An introduction to statistical learning. Vol. 112. New York: springer, 2013.
● Reinforcement learning: “Learning what to do - how to map situations to
actions, so as to maximize a numerical reward signal.”
○ Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018.
21
M2Lab
Major categories in ML
Discuss with your neighboring classmates and come up with an example for
each of the categories below. [10 minutes]
● Supervised learning: “Building a statistical model for predicting or estimating
an output based on one or more imputes.”
○ James, Gareth, et al. An introduction to statistical learning. Vol. 112. New York: springer, 2013.
● Unsupervised learning: “A statistical model built to learn the relationships and
structure without the supervising output.”
○ James, Gareth, et al. An introduction to statistical learning. Vol. 112. New York: springer, 2013.
● Reinforcement learning: “Learning what to do - how to map situations to
actions, so as to maximize a numerical reward signal.”
○ Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018.
22
