Towards a Sound Theory of Adaptation for the Simple Genetic Algorithm

The pace of progress in the fields of Evolutionary Computation and Machine Learning is currently limited -- in the former field, by the improbability of making advantageous extensions to evolutionary algorithms when their capacity for adaptation is poorly understood, and in the latter by the difficulty of finding effective semi-principled reductions of hard real-world problems to relatively simple optimization problems. In this paper we explain why a theory which can accurately explain the simple genetic algorithms remarkable capacity for adaptation has the potential to address both these limitations. We describe what we believe to be the impediments -- historic and analytic -- to the discovery of such a theory and highlight the negative role that the building block hypothesis (BBH) has played. We argue based on experimental results that a fundamental limitation which is widely believed to constrain the SGAs adaptive ability (and is strongly implied by the BBH) is in fact illusionary and does not exist. The SGA therefore turns out to be more powerful than it is currently thought to be. We give conditions under which it becomes feasible to numerically approximate and study the multivariate marginals of the search distribution of an infinite population SGA over multiple generations even when its genomes are long, and explain why this analysis is relevant to the riddle of the SGAs remarkable adaptive abilities.

Two Remarkable Computational Competencies of the Simple Genetic Algorithm

Since the inception of genetic algorithmics the identification of computational efficiencies of the simple genetic algorithm (SGA) has been an important goal. In this paper we distinguish between a computational competency of the SGA--an efficient, but narrow computational ability--and a computational proficiency of the SGA--a computational ability that is both efficient and broad. Till date, attempts to deduce a computational proficiency of the SGA have been unsuccessful. It may, however, be possible to inductively infer a computational proficiency of the SGA from a set of related computational competencies that have been deduced. With this in mind we deduce two computational competencies of the SGA. These competencies, when considered together, point toward a remarkable computational proficiency of the SGA. This proficiency is pertinent to a general problem that is closely related to a well-known statistical problem at the cutting edge of computational genetics.

The Fundamental Problem with the Building Block Hypothesis

Skepticism of the building block hypothesis (BBH) has previously been expressed on account of the weak theoretical foundations of this hypothesis and the anomalies in the empirical record of the simple genetic algorithm. In this paper we hone in on a more fundamental cause for skepticism--the extraordinary strength of some of the assumptions that undergird the BBH. Specifically, we focus on assumptions made about the distribution of fitness over the genome set, and argue that these assumptions are unacceptably strong. As most of these assumptions have been embraced by the designers of so-called competent genetic algorithms, our critique is relevant to an appraisal of such algorithms as well.