Training of Feed-Forward Neural Networks for Pattern-Classification Applications Using Music Inspired Algorithm

There have been numerous biologically inspired algorithms used to train feed-forward artificial neural networks such as generic algorithms, particle swarm optimization and ant colony optimization. The Harmony Search (HS) algorithm is a stochastic meta-heuristic that is inspired from the improvisation process of musicians. HS is used as an optimization method and reported to be a competitive alternative. This paper proposes two novel HS-based supervised training methods for feed-forward neural networks. Using a set of pattern-classification problems, the proposed methods are verified against other common methods. Results indicate that the proposed methods are on par or better in terms of overall recognition accuracy and convergence time. Keywords-harmony search; evolutionary methods; feed-forward neural networks; supervised training; pattern-classification