A Fireworks Algorithm for Single Objective Big Optimization of Signals

This paper presents a novel adaptation of the Fireworks Algorithm for single objective Big Data Optimization problems. In this context, the developed Single Objective Fireworks Algorithm (SOFWA) is proposed for solving the Big Optimization of Signals "Big-OPT" problem belonging to the Big Data Optimization problems class. Indeed, during an Encephalography record session, EEG signals are noised with artifacts coming from non-brain electric sources. the main purpose of the Big-OPT problem is to recover the true brain EEG signals and remove the maximum possible of artifacts. To this end, an optimization NP-Hard problem is defined. To solve it, SOFWA implements a modified search strategy to enhance the explorative capacities and increase the convergence speed of the original Fireworks Algorithm. To validate the performance of the proposed method, experiments have been performed over the Big-OPT EEG datasets. A comparison with recent state of the art approaches is also included. The study exhibits the competitive performance of the proposed method.