Scribd
Upload
6 views39 pages

Understanding Neural Networks Basics

This document discusses neural networks and their applications. It defines neural networks as artificial representations of the human brain that try to simulate its learning process. The key components of neural networks are described as the network architecture (how neurons are connected), setting the weights (training), and activation functions. Neural networks can perform supervised learning (with a desired output provided), unsupervised learning (without a desired output), and reinforcement learning (using a reward/punishment signal). Common applications of neural networks discussed include pattern recognition, prediction, control systems, and data processing.

Uploaded by

shradhdhey
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
6 views39 pages

Understanding Neural Networks Basics

This document discusses neural networks and their applications. It defines neural networks as artificial representations of the human brain that try to simulate its learning process. The key components of neural networks are described as the network architecture (how neurons are connected), setting the weights (training), and activation functions. Neural networks can perform supervised learning (with a desired output provided), unsupervised learning (without a desired output), and reinforcement learning (using a reward/punishment signal). Common applications of neural networks discussed include pattern recognition, prediction, control systems, and data processing.

Uploaded by

shradhdhey
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

April 2007

What is soft computing


It is

Types of Soft computing


Hard application.

What is Neural network


Neural network is artificial representation of human brain that tries to simulate its learning process.

DEFINITION OF NEURAL NETWORKS


operating in parallel whose

April 2007

BIOLOGICAL (MOTOR) NEURON

April 2007

NET

April 2007

ARTIFICIAL NEURAL NET


W1 X1 Y

W2 X2

The figure shows a simple artificial neural net with two input neurons (X1, X2) and one output neuron (Y). The inter connected weights are given by W1 and W2.
April 2007 8

April 2007

The neuron is

y (u b)
April 2007 10

MULTI LAYER ARTIFICIAL NEURAL NET

April 2007

11

OPERATION OF A NEURAL NET


x0

w0j w1j wnj Weighted m

Bias

x1
xn

f
Activation function

Output y

Input Weight vector x vector w

April 2007

12

BUILDING BLOCKS OF ARTIFICIAL NEURAL NET


Network Architecture (Connection between Neurons) Setting the Weights (Training)

Activation Function
April 2007 13

April 2007

14

LAYER PROPERTIES
Input Layer: Each input unit may be designated by an attribute value possessed by the instance.

Hidden Layer: Not directly observable, provides nonlinearities for the network. Output Layer: Encodes possible values.

April 2007

15

Learning in neural network


The main property of neural network is ability to learn. The learning is a process by which neurons in neural networks adapts itself to the stimulus by proper parameter adjustments resulting in production of desired response.

Classification of Learning in ANN


Parameter Learning. It updates connecting the weight in neural net. Structure Learning. It focuses on change in neural network structure.

Classification of Learning in ANN

Unsupervised Training - or wrong is provided.

April 2007

18

Supervised Learning
X Y Error (D-Y) signal D Desired O/P

Unsupervised Learning

X input

ANN W

Y Actual outpu

Reinforcement Learning
X input Neural Network W Error signal Error signal generator Y Actual output

R Reinforcement signal

ACTIVATION FUNCTION
Activation Function:-Is applied over net input to calculate net output.

ACTIVATION LEVEL )
Binary Sigmoidal activation function Bipolar Sigmoidal activation function

April 2007

22

ACTIVATION FUNCTION
Activation functions: (A)Identity (B)Binary step (C)Bipolar step (D)Binary sigmoidal (E)Bipolar sigmoidal

April 2007

23

SALIENT FEATURES OF ANN

April 2007

24

McCULLOCHPITTS NEURON
Neurons are sparsely and randomly connected

Firing state is binary (1 = firing, 0 = not firing)


All but one neuron are excitatory (tend to increase voltage of other cells) -One inhibitory neuron connects to all other neurons -It functions to regulate network activity (prevent too many firings)

April 2007

25

HEBB NETWORK
Donald Hebb stated in 1949 that in the brain, the learning is performed by the change in the synaptic gap. Hebb explained it:

When an axon of cell A is near enough to excite cell B, and repeatedly or permanently takes place in firing it, some growth process or metabolic change takes place in one or both the cells such that As efficiency, as one of the cells firing B, is increased.

April 2007

26

HEBB LEARNING
The weights between neurons whose activities are positively correlated are increased:

dw ij dt

~ correlatio n ( x i , x j )

Associative memory is produced automatically


The Hebb rule can be used for pattern association, pattern categorization, pattern classification and over a range of other areas.

April 2007

27

FEW APPLICATIONS OF NEURAL NETWORKS

April 2007

28

Model of multilayer perceptron

Back Propagation Learning

Input Layer computation

Hidden layer computation

Output layer computation

Kosko

BAM

Basic Coding Procedure of BAM

Retrieve the nearest of pattern pair, given any pair.

Energy Function of BAM

You might also like