Fast differential evolution for big optimization
Mohamed A. Meselhi, Saber Mohammed Elsayed, Daryl Essam, Ruhul Amin Sarker · 2017
Over the last few years, handling big optimization problems has become a key challenge for many domains, including, but not limited to, cyber security and healthcare systems, which generate a high volume of data, from different sources and requiring real-time processing. In the healthcare domain, dealing with captured Electroencephalographic (EEG) brain signals is vital. Although evolutionary algorithms have been successful in solving many complex optimization problems, limited work on using them for big optimization has been carried out. In addition, they are still computationally insufficient. In this paper, a fast differential evolution algorithm is designed to remove artifacts from EEG signals captured from non-brain sources, by using the parallel computing power of a graphics processing unit. The algorithm is tested on problems with 1024 decision variables, with and without noise, with the results showing the capability of the proposed algorithm of achieving high-quality solutions and reducing the computational time by 81%, compared with state-of-the-art algorithms.