Optimization of artificial neural net by genetic

Dissertant

Alyawi, Suhaib

University

University of Technology

Faculty

-

Department

Computer Sciences Department

University Country

Iraq

Degree

Master

Degree Date

2002

English Abstract

Artificial Neural Network or simply "neural network" attempts to achieve good performance via dense interconnection of simple computational elements, which called neurons or nodes, that organized in layers formed the neural network, There are many different kinds of learning rules, one of these is delta learning rule which car.

Be regarded as one of die most common ANN learning method.

Delta Rule minimizes the error between target value and actual value with gradient descent method, back propagation.

One of the drawbacks of backpropagati on is the initial values of the connection weight.

Recently, many researchers tried to use a hybrid technique (Genetic Algorithm with Rack propagation) in order to overcome the limitations of learning method.

Different sizes of ANN have been tried for pattern recognition.

It was shown that GAs provides good solutions, especially for small networks.

The problem of selecting the particular architecture of an ANN for a certain application is one of the primary difficulties in ANNs designing In this work, the ability of using GA to optimize the architecture of the ANNs has been measured.

GAs have been proved to be a good, and simple tool for optimizing.

The time that is required to train ANN using GA is half time needed to train the same ANN using classical method, and the time required to find the optimal structure of ANN using GA is little than the time that is required to find it using classical method.

Main Subjects

Information Technology and Computer Science

Topics

• Biology
• Embryology
• Genetics
• Chromosomes
• Biological adaptation
• Neural networks(Computer science)

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-305782