Investigations in multimodal optimization using differential evolution and its applications

Abstract

Multimodal Optimization algorithms are used to provide multiple solutions with the same or similar objective value. The need for multiple solutions arises in many real-life problems, such as neural network architecture search, data mining, pattern mining, and vehicle routing. There exist evolutionary algorithms which have tried to optimize and provide solutions to optimization problems. Most of them are singleobjective/ single-solution algorithms. There are multimodal optimization algorithms developed with the help of evolutionary algorithms, which exhibit the advantage of providing multiple choices to the user and selecting the optimal solution. Thus, it provides users with the flexibility of selecting different solutions under different circumstances. This thesis mainly investigates to design and develop the differential evolutionbased multimodal optimization algorithms for single and multiple objective(s) and their applications. To provide multiple optimal solutions, novel multimodal optimization algorithms are designed. First, we have proposed a multimodal optimization algorithm based on differential evolution by incorporating a niching technique.