Genetic Algorithms and its use with back-propagation network

TL;DR

This paper explores the use of genetic algorithms to enhance the training process of back-propagation neural networks, proposing a novel method for training set selection and an extension to traditional genetic algorithms.

Contribution

It introduces a new approach combining genetic algorithms with back-propagation networks for improved training set selection and extends the genetic algorithm methodology.

Findings

Enhanced training set selection improves neural network performance.

The proposed method reduces training time compared to traditional approaches.

New genetic algorithm extension offers better convergence properties.

Abstract

Genetic algorithms are considered as one of the most efficient search techniques. Although they do not offer an optimal solution, their ability to reach a suitable solution in considerably short time gives them their respectable role in many AI techniques. This work introduces genetic algorithms and describes their characteristics. Then a novel method using genetic algorithm in best training set generation and selection for a back-propagation network is proposed. This work also offers a new extension to the original genetic algorithms