I.J. Information Technology and Computer Science, 2013, 11, 32-41
Published Online October 2013 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijitcs.2013.11.04
Copyright © 2013 MECS I.J. Information Technology and Computer Science, 2013, 11, 32-41
A Novel Cat Swarm Optimization Algorithm for
Unconstrained Optimization Problems
Meysam Orouskhani
Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran Iran
E-mail: orouskhani@ce.sharif.edu
Yasin Orouskhani
BSc Student, Computer School of Sharif University of Technology, Tehran, Iran
E-mail: orouskhani@gmail.com
Mohammad Mansouri
Intelligent System Laboratory (ISLAB), Electrical and Computer engineering department, K.N.Toosi University
E-mail: mohammad_mansouri@ee.kntu.ac.ir
Mohammad Teshnehlab
Industrial Control Center of Excellence, Faculty of Electrical and Computer Engineering, K.N. Toosi University
E-mail: teshnehlab@ee.kntu.ac.ir
Abstract Cat Swarm Optimization (CSO) is one of
the new swarm intelligence algorithms for finding the
best global solution. Because of complexity, sometimes
the pure CSO takes a long time to converge and cannot
achieve the accurate solution. For solving this problem
and improving the convergence accuracy level, we
pr
adaptive inertia weight to velocity equation and then
use an adaptive acceleration coefficient. Second, by
using the information of two previous/next dimensions
and applying a new factor, we reach to a new position
update equation composing the average of position and
velocity information. Experimental results for six test
functions show that in comparison with the pure CSO,
the proposed CSO can takes a less time to converge and
can find the best solution in less iteration.
Index Terms Swarm Intelligence, Cat Swarm
Optimization, Evolutionary Algorithms
I. Introduction
Function Optimization is one of the important fields
in the computational intelligence theories. There are
many algorithms to find the global and local solutions
of the problems. Some of these optimization algorithms
were developed based on swarm intelligence. These
model into algorithm, such as Ant Colony Optimization
(ACO) which imitates the behavior of ants
[1]-[6]
, Particle
Swarm Optimization (PSO) which imitates the behavior
of birds
[2]
, Bee Colony Optimization (BCO) which
imitates the behavior of bees
[3]
and the recent finding,
Cat Swarm Optimization (CSO) which imitates the
behavior of cats
[4]
.
By simulating the behavior of cats and modeling into
two modes, CSO can solve the optimization problems.
In the cases of functions optimization, CSO is one of
the best algorithms to find the global solution. In
comparison with other heuristic algorithms such as PSO
and PSO with weighting factor
[7]
, CSO usually
achieves better result. But, because of algorithm
complexity, solving the problems and finding the
optimal solution may take a long process time and
sometimes much iteration is needed.
So in this article, we propose an improved CSO in
order to achieve the high convergence accuracy in less
iteration. First we use an adaptive inertia weight and
adaptive acceleration coefficient. So, the new velocity
update equation will be computed in an adaptive
formula. Then, our aim is to consider the effect of
previous/next steps in order to calculate the current
pos
Finally, we use an average form of position update
equation composing new velocity and position
information. Experimental results for standard
optimization benchmarks indicate that the proposed
algorithm rather than pure CSO can improve
performance on finding the best global solution and
achieves better accuracy level of convergence in less
iteration.
