Effectiveness of firefly algorithm based neural network in time series forecasting
Global optimization techniques such as Particle Swarm Optimizers (PSO) and Genetic Algorithm (GA) are now widely used for training Artificial Neural Networks (NN), particularly in time series forecasting problems. Firefly algorithm (FA) is a relatively new addition to the family of population based optimization technique that has shown promising result in a number of problems. In this work, we evaluate the effectiveness of FA trained NN in time series forecasting. In the experiments, three well known time series were used to evaluate the performance. Results obtained were compared with results from both PSO and Resilient Propagation (RPROP) trained NNs. FA based NN performed very well in forecasting all the time series considered, outperforming the bench-marks in two out of the three problems.
Keywords: Time series, Artificial Neural Network, Firefly Algorithm, Particle Swarm Optimization, Overfitting