A Species Conserving Genetic Algorithm for Multimodal Function Optimization

Abstract

This paper introduces a new technique called species conservation for evolving paral-lel subpopulations. The technique is based on the concept of dividing the population into several species according to their similarity. Each of these species is built around a dominating individual called the species seed. Species seeds found in the current gen-eration are saved (conserved) by moving them into the next generation. Our technique has proved to be very effective in finding multiple solutions of multimodal optimiza-tion problems. We demonstrate this by applying it to a set of test problems, including some problems known to be deceptive to genetic algorithms.

A Rigorous Runtime Analysis for Quasi-Random Restarts and Decreasing Stepsize

“…λ is the population size, 1 in our case (but the complexity results hold for any value of λ). RTS [10,11] and SCGA [15] [23] (see text). Right: Results on the sine function in dimension 1, for RDS and WRDS, and different strategies for σ.…”

“…Please refer to the survey proposed in [23] (or to the original papers of course). A popular family of MMO algorithms is based on niching techniques: sharing [8]; clearing [22], including the modified clearing approach proposed in [23]; crowding [6], including deterministic [18] and probabilistic [19] versions; clustering [30]; species conserving genetic algorithms [16]; and finally, different restart strategies, to the recent state-of-the-art restart with increasing population size [1]. Several other works have been devoted to MMO outside the evolutionary community.…”

Effect of Spatial Locality on an Evolutionary Algorithm for Multimodal Optimization

“…Since then, researchers have proposed different genetic algorithms for multimodal optimization problems. In particular, crowding [3], fitness sharing [8], and speciation [9,19] techniques are the most popular techniques. They have also been integrated in differential evolution [17] and demonstrated promising results [18,10].…”

Solving multimodal problems via multiobjective techniques with Application to phase equilibrium detection

“…Prominent examples are crowding [4] and fitness sharing [5], and their successors. More recent approaches include, but are not limited to: UEGO [6], clearing [7], species conservation [8], clustering based niching [9], and cellular EA (CEA) [10]. Although there is no commonly accepted formal definition of what a niching method is (see [11]), most of these algorithms may be subsumed under the term niching EA.…”