یکی از جدیدترین ابزارهای تقویت یادگیری بر اساس بهینه ساز PSO

یکی از جدیدترین ابزارهای تقویت یادگیری بر اساس بهینه ساز PSO

نام مقاله به فارسی:

نام مقاله به انگلیسی:

یکی از جدیدترین ابزارهای تقویت یادگیری بر اساس بهینه ساز PSO

یکی از جدیدترین ابزارهای تقویت یادگیری بر اساس بهینه ساز PSO

Developing an effective memetic algorithm that integrates the Particle Swarm Optimization (PSO) algo-rithm and a local search method is a difficult task. The challenging issues include when the local searchmethod should be called, the frequency of calling the local search method, as well as which particle shouldundergo the local search operations. Motivated by this challenge, we introduce a new ReinforcementLearning-based Memetic Particle Swarm Optimization (RLMPSO) model. Each particle is subject to fiveoperations under the control of the Reinforcement Learning (RL) algorithm, i.e. exploration, convergence,high-jump, low-jump, and fine-tuning. These operations are executed by the particle according to theaction generated by the RL algorithm. The proposed RLMPSO model is evaluated using four uni-modal andmulti-modal benchmark problems, six composite benchmark problems, five shifted and rotated bench-mark problems, as well as two benchmark application problems. The experimental results show thatRLMPSO is useful, and it outperforms a number of state-of-the-art PSO-based algorithms.

Memetic algorithmParticle Swarm OptimizationReinforcement learningLocal search

INTRODUCTION

Memetic-based optimization algorithms have been used suc-cessfully in many applications, e.g. DNA sequence compression [1],flow shop scheduling [2], multi-robot path planning [3], wirelesssensor networks [4], finance applications [5], image segmentation[6], and radar applications [7]. The main objective of developingmemetic-based algorithms is to exploit the benefits of both globaland local search methods and combine them into a single model. Asan example, the Particle Swarm Optimization (PSO) algorithm is aneffective global optimizer, and has been integrated with differentlocal search methods to produce a number of memetic PSO-basedmodels [1,2,8–۱۱]. The resulting models combine the global searchstrength of PSO and the refinement capability of local search meth-ods into a unified framework.

سایت ematlab.com سایت ایمتلب سایت ematlab.com سایت ایمتلب سایت ematlab.com سایت ایمتلب سایت ematlab.com سایت ایمتلب سایت ematlab.com سایت ایمتلب