Artificial neural network learning enhancement using Artificial Fish Swarm Algorithm
Shafaatunnur Hasan, Swee Quo Tan, Siti Mariyam Shamsuddin, Roselina Salleh Sallehuddin · Universiti Utara Malaysia Institutional Repository (Universiti Utara Malaysia) · 2011
Artificial Neural Network (ANN) is a new information processing system with large quantity of highly interconnected neurons or elements processing parallel to solve problems.Recently, evolutionary computation technique, Artificial Fish Swarm Algorithm (AFSA) is chosen to optimize global searching of ANN.In optimization process, each Artificial Fish (AF) represents a neural network with output of fitness value.The AFSA is used in this study to analyze its effectiveness in enhancing Multilayer Perceptron (MLP) learning compared to Particle Swarm Optimization (PSO) and Differential Evolution (DE) for classification problems.The comparative results indeed demonstrate that AFSA show its efficient, effective and stability in MLP learning.