Design of Computational Intelligence Techniques for Solving Complex Engineering Optimization Problems

Abstract

Due to the lack of effective strategies to accurately resolve the sen- newlinetiment analysis problems and identify fake reviews, there is still newlinescope for developing a practical sentiment analysis algorithm. Fea- newlineture selection is a technique commonly used in Data Mining and newlineMachine Learning. Traditional feature selection methods, when ap- newlineplied to large datasets, generate a large number of feature subsets. newlineSelecting optimal features within this high-dimensional data space newlineis time-consuming and negatively affects the system s performance. newlineTherefore, during this research work, a detailed study of nature- newlineinspired algorithms and their usability in engineering optimization newlineproblems was carried out. Two recently developed nature-inspired newlinealgorithms, namely spider monkey optimization (SMO) and salp newlineswarm algorithm (SSA), are considered for this research. The SMO newlineand SSA are competitive swarm intelligence-based algorithms to newlineget rid of complex real-life optimization problems as the optima newlinesearch process of SMO is a little bit biased by the random com- newlineponent that drives it with high explorative searching steps. The newlinenovel contribution of this research includes two new variants of newlineSMO and one new variant of SSA. SMO variants are named hybrid newlinespider monkey optimization (HSMO) and memetic spider monkey newlineoptimization (MeSMO). HSMO was applied for Twitter sentiment newlineanalysis, and MeSMO was used to solve spam review detection newlineproblems. The third variant is the binary salp swarm algorithm newline(bSSA) employed for feature selection with the hybrid data trans- newlineformation approach. newline