Chapter 3 :
Optimal Estimate Training
For OET, the activation function f( ) should be a continuous invertible
function such as the hyperbolic tangent (tanh).
The number of processing elements in the input layer, n, is the same as the
number of processing
% Class 05c Dr. G. Pang Sept, 2013
% Demo of self-organizing feature map (SOFM) Euclidean Distance
% for clustering of a 7-D data
clear all, close all
% 1. Form the data, there are 5 clusters, 20 in each cluster
% S : 0 - 0.4 ; M : 0.45 - 0.65 ; L :
Class 03b: ANN Assignment on
Number Recognition
Objective: Design of Back-Propagation ANN
for the recognition of the ten numbers/digits.
You should write clearly how you design your
network, obtain results, carry out the
evaluation and analyze the perfo
% Class 04d: ELEC 6604 Dr. G. Pang Sept, 2013
% Demo of self-organizing feature map (SOFM)
% Suppose we have 100 samples of a 7-D data
% What can we say about the data ?
xdata =
0.9612 0.8648 0.0222 0.3751 0.4906 0.0989 0.6135
0.5009 0.3997 0.5593 0.
% Tutorial_ColorImageSeg_Perceptron with EXERCISE
% Copyright by Dr. G. Pang
% October 2, 2012
% Perceptron Demo : Color Image Segmentation
%
% Description: Different colors are defined and a perceptron
% is trained for the color image segmentation proble
Class 02c DemoB
More on the Artificial Neural Network
Why do we need to sum the squared errors?
1 q k
E (c j c j ) 2
2 j 1
The sum of the squared errors is a useful indicator of the
networks performance. The BP training algorithm
attempts to minimize this
Class 03a : Applications of
Neural Network Technology
What is a neural network ?
A neural network can be defined as a model of
reasoning based on the human brain. It is a very
popular approach to machine learning.
It represents a class of very powerful,
Class 05a :
Radial Basis Function Networks
A radial basis function (RBF) network is a two-layer network
whose output units form a linear combination of the basis (kernel)
functions computed by the hidden units.
The basis functions in the hidden layer prod
Class 05b:
Hopfield Neural Network
The Hopfield network emulates the
human memorys
associative characteristics.
It is a kind of recurrent neural network.
1
The Hopfield Network
A recurrent neural network has feedback loops from its
outputs to the inputs.
Class 01b: Introduction to ANN
Artificial Neural Networks
What is a neural network ?
A neural network can be defined as a
model of reasoning based on the human
brain. It is a very popular approach to
machine learning.
Computations in the human brain
As
ELEC 6604
Class 01a: Introduction to
Artificial Intelligence
January 11, 2017
Dr. Grantham Pang
[email protected]
Fundamental Questions of AI
Can machines think?
( I think, there I am. )
Is there mind without communication?
Is there language without thinki
% Class 04c: Tutorial_ColorImageSeg_SOFM.doc
%
% Dr. G. Pang Sept 28, 2013; updated Oct 6, 2015
% Demo of self-organizing feature map (SOFM) Euclidean Distance
% for Color Image Segmentation
clear all, close all
% 1. Form the data
%
img1=imread('3503A.bmp
Chapter 4:
Unsupervised Learning
Unsupervised Learning
Can a neural network learn without a teacher ?
The main property of ANN is an ability to learn from its
environment, and to improve its performance through
learning. BP and OET are supervised or activ
The most common basis function chosen is a Gaussian
function, in which case the activation level hj of hidden unit j
is calculated by
T
2
Chapter 5 :
Radial Basis Function Networks
h j = exp[ ( X C j ) ( X C j ) / 2 j ]
A radial basis function (RBF) netwo
Chapter 2 : Multilayer
Why do we need a hidden layer ?
Neural Networks
What is a multilayer neural network ?
A multilayer neural network is a feedforward neural network
with one or more hidden layers.
The network consists of an input layer of source neu
Fundamental Questions of AI
ELEC 3503
Introduction to
Artificial Intelligence
Can machines think?
Is there mind without communication?
Is there language without living?
Is there intelligence without life?
thinking
September 2, 2013
Dr. Grantham Pang
gpang
The Hopfield Network
Chapter 6:
Hopfield Neural Network
A recurrent neural network has feedback loops from its
outputs to the inputs. The presence of such loops has a
profound impact on the learning capability of the network.
How does the recurrent netw
M
S
x x
o o
B
o
o
x
o
o
o
# #
o
#
x
x
x #
o
B (x)
1+ e
1
- (x - 15)
A (x)
B (x)
a number more than
two but not large
o o
o
o
o o
A (x)
o
o
o o
1 1 0.5
+ +
1 2 3
0.5 1 1 1 1 0.5
+ + + + +
3 4 5 6 7 8
0.5 1 1
+ +
8 9 10
These are not arithmetic operator
0.6
NO
0.49
YES
YES
0.51
0.6
60%
40%
due to rule 1
due to rule 2
There are other ways of doing things:
e.g. simply set speed = 60%x1000+40%x500
= 800
Single value to represent the
resulting output speed
AI B
0 .2 + 0 .3
2
3
Note that if the algebraic product is used to define A I B
i.e. A I B ( x) = A ( x) B ( x), then A I B = 0.1 + 0.12
2
3
Smallest t-norm any t-norm largest t-norm
AUB
A = 0.6 + 0.5 + 0.3 B = 0.2 + 0.4 + 0.9 A U B = 0.6 + 0.5 + 0.4 +
COMP 328: Machine Learning
Lecture 2: Naive Bayes Classiers
Nevin L. Zhang
Department of Computer Science and Engineering
The Hong Kong University of Science and Technology
Spring 2010
Nevin L. Zhang (HKUST)
COMP 328
Spring 2010
1 / 34
Two dierent types o