Scribd
Upload
492 views

Adaline Madaline

ADALINE is an adaptive linear neuron that uses the Widrow-Hoff learning rule to minimize error. It can be used for applications like system identification, adaptive control, and market prediction. However, ADALINE has limitations in that it can only model linear functions and may get stuck in local minima during learning. To model nonlinear functions, multilayer networks called MADALINE were introduced that combine ADALINE units with nonlinear elements like AND gates.

Uploaded by

jain14prateek
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
492 views

Adaline Madaline

ADALINE is an adaptive linear neuron that uses the Widrow-Hoff learning rule to minimize error. It can be used for applications like system identification, adaptive control, and market prediction. However, ADALINE has limitations in that it can only model linear functions and may get stuck in local minima during learning. To model nonlinear functions, multilayer networks called MADALINE were introduced that combine ADALINE units with nonlinear elements like AND gates.

Uploaded by

jain14prateek
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 8

Neural Networks and Fuzzy Systems

ADALINE stands for an adaptive linear neuron


ADALINE has a linear activation function, i.e. f(net)=net;




Learning rule(Widrow-Hoff learning rule):



ADALINE
Learning objective:
net= d(X)
1 ) ( = '
j
net f
0 , ) 1 ( ) ( > + = c x c j w j w
i j i i
o
Neural Networks and Fuzzy Systems
Learning Algorithm
c
x
x
x
Neural Networks and Fuzzy Systems
ADLINE is a linear approximator. As a linear classifier, it
often connects with a hard-limiter at the output
ADALINE
Learning may fall into a local minimum, i.e. 0 but ; 0 = =
c
c
E
w
E
i
Neural Networks and Fuzzy Systems
Model identification of a linear system
Adaptive Control;
Market Prediction
Applications of ADALINE
Controller
Adaline Model
Plant
Controller Parameters
u
y
plant

u
c

... ) 1 ( ) (
... ) 2 ( ) 1 ( ) (
1
2 1
+ + +
+ + =

t u w t u w
t y w t y w t y
m m
Neural Networks and Fuzzy Systems
Stabilization of an invert pendulum by learning from
demonstration
Applications of ADALINE
... ) 2 ( ) 1 ( ) (
2 1
+ O + O = O t w t w t
ARM ARM C
Neural Networks and Fuzzy Systems
Adaptive Noise Cancellation :
Applications of ADALINE
Neural Networks and Fuzzy Systems
Can only deal with linear function approximation;
If an attempt is made to implement a non-linear function,
the solution that it arrives at will contain errors but will be a
kind of best fit, in the sense that it will be the best that can
be achieved using just a single layer NN;
In order to deal with non-linear approximation:
1. Introduce nonlinear high order terms for generating a
nonlinear separation line:
2. A solution is to have a layer of ADALINEs (with hard
limiter) feeding into a logic device such as an AND gate.
These multi-layered networks are called MADALINE.
ADALINE

+ =
ij i ijk i ik k
x x w x w y
Neural Networks and Fuzzy Systems
Summary
Adaline
Madaline
Readings
Pictons book: Chapter 2 (Adaline)
Haykins book: Chapter 4 (if you want to know convergence
proof of the learning law)
For the reading week
Widrow, 30 years of adaptive neural networs: perceptron,
madaline and backpropagation
Breazeal,Challenges in building robots that imitate people

You might also like