Certain investigations on the performance of face recognition system using classifiers and optimizers

Abstract

The role of face recognition gains persistence in the area of newlineinformation security, smart cards, aadhar cards, surveillance systems and newlineaccess control. The need for face recognition is the principal concern of the newlinemodern age. But the accuracy of the system in traditional methods is affected newlineby pose variation, illumination conditions and different face expressions. newlineHence there must be a necessitation of enhancing the mechanism in every newlineresearch within the discipline. To achieve this, optimization based newlineclassification techniques can be formulated and evaluated in this research so newlinethat to facilitate the recognition of faces in an effective way. newlineThis thesis is built on the premise of the efficiency of population newlinebased metaheuristic algorithms on classification techniques. The natural newlinechoice for the investigation fell on heuristic algorithms since they possess newlinemomentous attributes, intensification and diversification and they are capable newlineof suggesting near good solutions among the several obtainable solutions. newlineThepropositionofthethesis lies on thevalidation o f Tuned Bacterial Foraging newlinea lgorithm based on chemical attraction and repulsion, Intensified Firefly newlinealgorithm and Micro batch particle swarm optimization algorithm on support newlinevector machines, extreme learning machines and deep convolution neural newlinenetwork respectively newline newline