matejgrcic/neural-network-architecture-optimization

An algorithm for evolving architecture of artificial neural networks

Neural Network Architecture Optimization

Overview

This project implements an algorithm for evolving architecture of artificial neural networks. Genetic algorithms are used for evolving network's architecture. The direct coding scheme is used for representing networks architecture in each individual. The neural network is trained using PSO algorithm.

Links

• Finding Optimal Neural Network Architecture Using Genetic Algorithms
• Artificial Neural Network
• PSO Algorithm
• Genetic Algorithms

TODO - for v2

• Refactor the way of coding NN's architecture to GA individual
• Optimize performances
• Update project to Java 11