Enhancing Niching Method: A Learning Automata Based Approach
Mohsen Jahanshahi
1+
, Mohammad Sadegh Kordafshari
2
, Majid Gholipour
2
1
Department of Computer Engineering, Central Tehran Branch, Islamic Azad University, Tehran, Iran
2
Department of Computer Engineering, Qazvin branch, Islamic Azad University, Qazvin, Iran
Abstract. In some problems of mathematics there are many functions which have several optimum points
which all, should be computed. The method NichePSO is previously designed to accomplish this. This paper
presents a new memetic-based scheme to enhance the NichePSO. This work utilizes a powerful soft
computing tool namely Learning Automata as the local search algorithm in our proposed memetic approach.
Numerical results demonstrate the superiority of the proposed method over the original NichePSO in terms of
convergence and diversity.
Keywords: PSO, Niching, Learning automata
1. Introduction
Optimization is the minimization or maximization of an objective function that normally is done with
consideration of limitations identifying conditions of a problem. In other words, it means the finding of the
best solution for a given problem. Some problems have several local optimums which may be computed
using Niching method. In this paper, memetic method is used for increase of convergence speed and also the
increasing rate of particles’ diversity in Niching method, as two assessment factors of search methods. In this
research, learning automata is used as a local search algorithm in memetic method.
The rest of the paper is organized as follow: in section 2, PSO method is introduced. In section 3,
NichePSO is briefly discussed. Section 4 explains the local search methods which are used and then
proposed methods are discussed in section 5. The results of simulations are included in section 6. Section 7 is
conclusion.
2. Particle Swarm Optimization
Particle Swarm Optimization (PSO) was discussed in [1][2][3][4] and then has been improved for many
years (e. g. [5][6]). It can be also used for several applications such as numerical optimization. Its main idea
had been taken from Birds or fishes swarm behaviour which are searching for meal. Some birds are
searching for meal randomly. Only a piece of meal may be found in the mentioned space. None of them
knows about the real place of the meal. One of the best strategies is to follow a bird that is placed in
minimum distance to a meal. In fact, this strategy is basis of PSO algorithm. This method is an effective
technique for solution of optimization problems based on a swarm behaviour. In this method, each member
of a swarm is called a particle who attempts to achieve a final solution with adjustment of its route and
movement to the best personal and swarm experiences. In PSO algorithm, a solution is called a particle the
+
Corresponding author.
E-mail address: mjahanshahi@iauctb.ac.ir
2011 International Conference on Computer and Software Modeling
IPCSIT vol.14 (2011) © (2011) IACSIT Press, Singapore
174