Particle Swarm Optimization with Diversive Curiosity - An Endeavor to Enhance Swarm Intelligence
Hong Zhang, Masumi Ishikawa · IAENG international journal of computer science · 2008
How to manage trade-o between ex- ploitation and exploration in Particle Swarm Opti- mization (PSO) for eciently solving various opti- mization problems is an important issue. In order to prevent premature convergence in PSO search, this paper proposes a new method, Particle Swarm Optimization with Diversive Curiosity (PSO/DC). A key idea of proposed method is to introduce a mechanism of diversive into PSO for pre- venting premature convergence and for managing exploration-exploitation trade-o. Diversive curios- ity is represented by an internal indicator that detects marginal improvement of a swarm of particles for cer- tain number of iterations, and forces them to contin- ually explore an optimal solution to a given optimiza- tion problem. Applications of proposed method to a 2-dimensional optimization problem well demon- strate its eectiveness. Our experimental results in- dicate that performance (100%) by proposed method is superior in terms of success ratio to that (60%) by PSO model optimized by EPSO, and basically accord with finding called the zone of curiosity in psychology.