Artificial neural network

Artificial Neural Network

Introduction to Artificial Neural
Network
Models of the
brain and
nervous system

 An Artificial Neural Network is a computer program that can
recognize pattern in a given collection of data and produce a
model for that data.
 These are highly parallel means process information much
more like brain than a serial computer.
Introduction to Artificial Neural
Network

Inspiration from Neurobiology
The original inspiration for the term Artificial Neural Network came
from examination of Central Nervous Systems and their
Neurons
Axons
Dendrites
Synapses

An Artificial Neural Network
is composed of many
artificial neurons that are
linked together according to
a specific network
architecture. The objective
of the neural network is to
transform the inputs into
meaningful outputs.

Mathematical Representation
The neuron calculates a weighted sum of inputs and compares it to a
threshold. If the sum is higher than the threshold, the output is set to 1,
otherwise to -1.
Non-Linearity

How do Artificial Neural Network work?
Adaptive System
Input Output
Cost
Desired
Training algorithm
Error

Why we use Artificial Neural Network?
 Adaptive learning
 Self-Organization
 Real Time Operation
 Fault Tolerance via Redundant Information Coding

Learning Methods in Artificial Neural Network
Learning Process is categorized into three parts:
 Supervised Learning
 Unsupervised Learning
 Reinforced Learning

Supervised Learning
 A teacher is present during learning process and presents
expected output.
 Every input pattern is used to train the network.
 Learning process is based on comparison between the
calculated output and desired output.
 The error generated is used to change network parameters
that result improved performance.

Unsupervised Learning
 No teacher is present.
 The desired output is not presented to the network.
 The system learns of it own by discovering and adapting to
the structural features in the input patterns.

Reinforced Learning
 A teacher is present but does not present the expected output
to the network.
 A reward is given for correct output and a penalty for wrong
answer.

Where are ANN used?
 Recognizing and matching complicated, or in completed
pattern
 Data is unreliable

Applications of Artificial Neural Networks
 Function approximation
 Classification
 Data mining
 Time series prediction

Function Approximation
Function approximation:
Create continuous input output
map
Inputs Outputs
The ANN must approximate f() in order to find the appropriate output for
each set of inputs.

Example of Function Approximation
 Use a neural network to create a model that can be used
to estimate the body density(e.g. body fat)
Two steps:
 Train the network
 Use the network

Classification
 Similar to function approximation except that
the output is a class
Classification
Interpret output as
a class
Numerical
Inputs
Numerical
Outputs
On/off
Outputs

 Recognition
 Pattern recognition
 Character recognition
 Handwriting: processing checks
 Data association
 Not only identify the characters that were scanned
but identify when the scanner is not working properly

 Data Conceptualization
 infer grouping relationships
e.g. extract from a database the names of those most
likely to buy a particular product.
 Data Filtering
 Planning
 Unknown environments
 Sensor data is noisy
 Fairly new approach to planning

Types of Artificial Neural Networks
They can be distinguished by:
 Their type
 Their structure
 The learning algorithm they
use

Types of Artificial Neural Networks
 Perceptron
 Multi-Layer- Perceptron
 Backpropagation Net
 Hopfield Net
 Kohonen Feature Map

Perceptron
Perceptron structure

Multi-Layer-Perceptron
Multi-Layer-
Perceptron
structure

Back propagation Net
Backpropagation
Net structure

Hopfield Net
Hopfield Net
structure

Kohonen Feature Map
Kohonen Feature Map
structure

Artificial neural network

More Related Content

What's hot (20)

Similar to Artificial neural network (20)

Recently uploaded (20)

Artificial neural network