Cathy Escazut

An Evolutionary Approach for Time Dependent Optimization

Many real-world problems involve measures of objectives that may be dynamically optimized. The application of evolutionary algorithms, such as genetic algorithms, in time dependent optimization is currently receiving growing interest as potential applications are numerous ranging from mobile robotics to real time process command. Moreover, constant evaluation functions skew results relative to natural evolution so that it has become a promising gap to combine effectiveness and diversity in a genetic algorithm. This paper features both theoretical and empirical analysis of the behavior of genetic algorithms in such an environment. We present a comparison between the effectivenss of traditional genetic algorithm and the dual genetic algorithm which has revealed to be a particularly adaptive tool for optimizing a lot of diversified classes of functions. This comparison has been performed on a model of dynamical environments which characteristics are analyzed in order to establish the basis of a testbed for further experiments. We also discuss fundamental properties that explain the effectiveness of the dual paradigm to manage dynamical environments.

Genetic operators in a dual genetic algorithm

It is not clear that the current distinction between crossover and mutation is necessary. We show that it is possible to implement one and only one general operator which can specialize crossover or mutation operators. We investigate this alternative. Our approach consists in inserting doubles in the population of chromosomes. This article argues that explicit mutations are unnecessary. Indeed, in dGAs without a mutation operator, chromosomes undergo the mutation effect. The dual genetic search provides a source of power for searching in a changing environment. Within this paper, a first effort is presented towards incorporating the feature of self-adaptation into GAs by using adaptive mutation rates. Finally, we study the effects of explicit mutation on a dual search space. We show that a contraction of the Hamming distance is induced from mutation. As a consequence, a dGA allows to increase the capabilities of evolution on rugged fitness landscapes.

Deceptiveness and neutrality the ND family of fitness landscapes

When a considerable number of mutations have no effects on fitness values, the fitness landscape is said neutral. In order to study the interplay between neutrality, which exists in many real-world applications, and performances of metaheuristics, it is useful to design landscapes which make it possible to tune precisely neutral degree distribution. Even though many neutral landscape models have already been designed, none of them are general enough to create landscapes with specific neutral degree distributions. We propose three steps to design such landscapes: first using an algorithm we construct a landscape whose distribution roughly fits the target one, then we use a simulated annealing heuristic to bring closer the two distributions and finally we affect fitness values to each neutral network. Then using this new family of fitness landscapes we are able to highlight the interplay between deceptiveness and neutrality.

A Minimal Model of Communication for a Multi-agent Classifier System

Classifier systems are rule-based systems dedicated to the learning of more or less complex tasks. They evolve toward a solution without any external help.When the problem is very intricate it is useful to have different systems, each of them being in charge with an easier part of the problem. The set of all the entities responsible for the resolution of each sub-task, forms a multi-agent system. Agents have to learn how to exchange information in order to solve the main problem. In this paper, we define the minimal requirements needed by a multi-agent classifier system to evolve communication. We thus design a minimal model involving two classifier systems which goal is to communicate with each other. A measure of entropy that evaluates the emergence of a common referent between agents has been finalised. Promising results let think that this work is only the beginning of our ongoing research activity.

Learning Disjunctive Normal Forms in a Dual Classifier System (Extended Abstract)

Genetics-Based Machine Learning systems suffer from many problems as representational weaknesses. We propose to introduce more general structures we used to learn disjunctive normal forms. Results show how our model can be used to discover and maintain complete classifier solutions.

Dynamic Management of the Specificity in Classifier Systems

The estimation of the rule usefulness in a classifier system is faced to the creditapportionment problem. Usually, the apportionning of payoffs process is performed by the bucket brigade algorithm. However, some works have shown that this algorithm presents some difficulties.