Particle Swarm Optimization (PSO) with Stochastic Data

Abstract

In photonics, numerous phenomena display stochastic behavior for example phase noise in LASERS or speckle in imaging. The study on stochastic data remains as a strong subject of interest. Evolutionary computational techniques serve as a vehicle for understanding and analyzing such chaotic environments. In this work Particle Swarm Optimization (PSO) algorithm, a popular evolutionary computational technique is extended to handle one such stochastic data. PSO is a popular optimization technique inspired from observing the natural optimization processes involving flocking of birds or swarming of insects especially bees. The idea of achieving a near target using individual and collective intelligence is the main feature behind the development of this optimization technique. The technique uses evolutionary process to search and achieve a near optimal solution. Through this research study we have made an attempt to comprehend the application of PSO Algorithm to chaotic environment where stochastic data play an integral role in affecting the PSO performance. This has not been attempted yet since the PSO has been applied mostly to deterministic problems. Numerical experiments were performed to analyze and predict the behavior of the stochastic datasets. In this research study, PSO algorithm is used along with a combination of Technical Indicators to investigate the stochastic patterns exhibited in foreign exchange (forex) market. Main Contribution of the research is to explore the possibilities of using PSO as a bridge to understand and comprehend chaotic environments for fuelling further research and additionally applying PSO as an independent tool for designing and developing better future photonic devices that presently suffer from effects of stochastic behavior. A foreign exchange market data serve as a test-bed for the PSO studies.