Study of PSO and Firefly algorithm based Feed-forward neural network training algorithms

Over the past few years, engineered neural network training algorithms based on nature-inspired algorithms have demonstrated their efficacy in proving their dominance over many traditional algorithms. Using the population-based approach has significantly increased the accuracy of the neural artificial network and converged towards higher accuracy. These approaches are well known to solve various problems in engineering and science. This paper discusses the algorithm and flowchart of the entire NI-algorithm hybridization process and the standard training algorithm for the neural network. The firefly and PSO optimization algorithms are considered in this work for analysis purpose. Such algorithms have different parameters and are therefore also discussed how to correct the parameters for that algorithm. UCI machine learning data set is finally tested for the efficiency of these algorithms. To find the best, the comparative analysis is provided on the results of the above-mentioned algorithms.