Optimization of Connecting Weights of an Artificial Neural Network Using Simple Particle Swarm Optimization Technique and its Variants

Abstract

Learning is one of the major parameters when dealing with artificial neural networks. It is considered to be one of the most difficult tasks in machine learning. Slow convergence speed is one of the major drawbacks which was seen while training the ANN with a conventional technique. Artificial neural networks come under the field of artificial intelligence. As ANNs play an important role in solving many complex problems in real-time engineering problems related to various sectors, enormous research is carried out throughout the globe. The learning can be enhanced by improving the connecting weights of an artificial neural network. The particle swarm optimization algorithm, which comes under the category of swarm intelligence, is the most popular technique used in today's research. The paper addressed optimizing the connection weights of an ANN using a swarm intelligence technique called Particle swarm optimization (PSO) besides its variants. It was simulated in the MATLAB domain and the results have been compared among the simple PSO and its variants.