Structural Feature Based Anomaly Detection for Packed Executable Identification - Computational Intelligence in Security for Information Systems - page 293

A New Particle Swarm Optimization for Dynamic
Environments
Hamid Parvin 1 , Behrouz Minaei 1 , and Sajjad Ghatei 2
1 School of Computer Engineering, Iran University of Science and Technology (IUST),
Tehran, Iran
{parvin,minaei}@iust.ac.ir
2 Department of Computer Engineering, Islamic Azad University Ahar Branch,
Young Researchers Club, Ahar, Iran
[e-mail]
Abstract. Dynamic optimization in which global optima and local optima
change over time is always a hot research topic. It has been shown that particle
swarm optimization works well facing into dynamic environments. From another
hands, learning automata is considered as an intelligent tool (agent) which can
learn what action is the best one interacting with its environment. The great
deluge algorithm is also a search algorithm applied to optimization problems. All
these algorithms have their special drawbacks and advantages. In this paper it is
examined can the combination of these algorithms results in the better
performance dealing with dynamic problems. Indeed a learning automaton is
employed per each particle of the swarm to decide whether the corresponding
particle updates its velocity (and consequently its position) considering the best
global particle, the best local particle or the combination global and local
particles. Water level in the deluge algorithm is used in the progress of the
algorithm. Experimental results on different dynamic environments modeled by
moving peaks benchmark show that the combination of these algorithms
outperforms Particle Swarm Optimization (PSO) algorithm, Fast Multi-Swarm
Optimization (FMSO) method, a similar particle swarm algorithm for dynamic
environments, for all tested environments.
Keywords: Particle Swarm Optimization, Great Deluge, Learning Automaton,
Moving Peaks, Dynamic Environments.
1 Introduction
The standard particle swarm optimization algorithms have been performed well
for static environment. Also it is shown that original PSO can't handle dynamic
environments. So researchers turn to new variations of PSO to overcome its
inefficiency [1].
Hu and Eberhart proposed re-randomization PSO (RPSO) for optimization in
dynamic environments [13] in which some particles randomly are relocated after a
change is detected or when the diversity is lost, to prevent losing the diversity. Li and
Dam [14] showed that a grid-like neighborhood structure used in FGPSO [11] can