Distribution of populations generated by multi-objective optimization evolutionary algorithms

• Autorzy: Dziemieszkiewicz P.
• Konferencja: Evolutionary Computation and Global Optimization 2006 / National Conference (9 ; 31.05-2.06.2006 ; Murzasichle, Poland)
• Języki publikacji: EN

Abstrakty

EN

This papers considers the distribution of population generated by multi-objective evolutionary algorithms. Some algorithms have a tendency to produce multimodal distributions of populations while other algorithms produce a compact cluster of solutions. Particular MOEAs have a tendency to one of these distributions. This paper will show that one feature of an algorithm defines it - the selection method. If selection is based on the evolution of parents, the algorithm generates a compact population every time. When the selection is based on the fitness of the offspring, the algorithm is able to generate bimodal distribution of population. Five MOEA algorithms were analyzed : VEGA, two types of tournament and two methods proposed by author (MOEA with competitive selection and MOEA with protective selection). The results of the simulation confirmed that, depending on the selection method selected, the analyzed algorithms generate the expected distribution of population. This paper does not assess which algorithms is better or worse but merely seeks to explain the reason for the algorithms' properties.

Słowa kluczowe

EN

evolutionary algorithm; EA; multi-objective optimization evolutionary algorithm; MOEA; distribution of populations; multi-objective optimization

PL

algorytm ewolucyjny; wielokryterialny algorytm ewolucyjny; rozkład populacji; optymalizacja wielokryterialna

• Wydawca: Oficyna Wydawnicza Politechniki Warszawskiej
• Czasopismo: Prace Naukowe Politechniki Warszawskiej. Elektronika
• Rocznik: 2006
• Tom: z. 156
• Strony: 117--124
• Opis fizyczny: Bibliogr. 7 poz., tab., wykr.

Twórcy

autor

Dziemieszkiewicz P.

• Wrocław University of Technology, Institute of Computer Engineering, Control and Robotics, Wrocław, Poland,

Bibliografia

• [1] C.A. Coello Coello, D.A. Veldhuizen, G.B. Lamont: Evolutionary Algorithms for Solving Multi-Objective Problems. Kluwer Academic Publishers, New York, 2002.
• [2] A.K. De Jong: An Analysis of the Behavior of a Class of Genetic Adaptive Systems. PhD thesis, University of Michigan.
• [3] P. Dziemieszkiewicz, R. Galar: Simulation of processes of population diversification in heterogeneous environment - in Polish, X Conference - Simulation of dynamic processes, Zakopane, 1998.
• [4] C. Fonseca, P. Fleming: An Overview of Evolutionary Algorithms in Multiobjective Optimization, Fvolutionary Computation, Vol. 3, No. 3, 165-180, 1995.
• [5] D.E. Goldberg, J. Richardson: Genetic Algorithms with sharing for multimodal function optimization. red. Grefenstette, 41-49, 1987.
• [6] J.D. Schaffer: Multiple Objective Optimization with Vector Evaluated Genetic Algorithm. In Genetic Algorithm and their Applications: Proceedings of the First International Conference on Genetic Algorithms, Hillsdale, New Jersey, 93-100, 1985.
• [7] http://www.lania.mx/~ccoello/EMOO/EMOObib.html - extensive MOEA bibliography.