Anna I. Esparcia-Alcázar

Bloat control operators and diversity in genetic programming: A comparative study

This paper reports a comparison of several bloat control methods and also evaluates a recent proposal for limiting the size of the individuals: a genetic operator called prune and plant. The aim of this work is to test the adequacy of this method. Since a preliminary study of the method has already shown promising results, we have performed a thorough study in a set of benchmark problems aiming at demonstrating the utility of the new approach. Prune and plant has obtained results that maintain the quality of the final solutions in terms of fitness while achieving a substantial reduction of the mean tree size in all four problem domains considered. In addition, in one of these problem domains, prune and plant has demonstrated to be better in terms of fitness, size reduction, and time consumption than any of the other bloat control techniques under comparison. The experimental part of the study presents a comparison of performance in terms of phenotypic and genotypic diversity. This comparison study can provide the practitioner with some relevant clues as to which bloat control method is better suited to a particular problem and whether the advantage of a method does or does not derive from its influence on the genetic pool diversity.

A Genetic Programming Approach for Bankruptcy Prediction Using a Highly Unbalanced Database

In this paper we present the application of a genetic programming algorithm to the problem of bankruptcy prediction. To carry out the research we have used a database of Spanish companies. The database has two important drawbacks: the number of bankrupt companies is very small when compared with the number of healthy ones (unbalanced data) and a considerable number of companies have missing data. For comparison purposes we have solved the same problem using a support vector machine. Genetic programming has achieved very satisfactory results, improving those obtained with the support vector machine.

Testing the Intermediate Disturbance Hypothesis: Effect of Asynchronous Population Incorporation on Multi-Deme Evolutionary Algorithms

In P2P and volunteer computing environments, resources are not always available from the beginning to the end, getting incorporated into the experiment at any moment. Determining the best way of using these resources so that the exploration/exploitation balance is kept and used to its best effect is an important issue. The Intermediate Disturbance Hypothesis states that a moderate population disturbance (in any sense that could affect the population fitness) results in the maximum ecological diversity. In the line of this hypothesis, we will test the effect of incorporation of a second population in a two-population experiment. Experiments performed on two combinatorial optimization problems, MMDP and P-Peaks , show that the highest algorithmic effect is produced if it is done in the middle of the evolution of the first population; starting them at the same time or towards the end yields no improvement or an increase in the number of evaluations needed to reach a solution. This effect is explained in the paper, and ascribed to the intermediate disturbanceproduced by first-population immigrants in the second population.

Fitness approximation for bot evolution in genetic programming

Estimating the fitness value of individuals in an evolutionary algorithm in order to reduce the computational expense of actually calculating the fitness has been a classical pursuit of practitioners. One area which could benefit from progress in this endeavour is bot evolution, i.e. the evolution of non-playing characters in computer games. Because assigning a fitness value to a bot (or rather, the decision tree that controls its behaviour) requires playing the game, the process is very costly. In this work, we introduce two major contributions to speed up this process in the computer game Unreal Tournament 2004™. Firstly, a method for fitness value approximation in genetic programming which is based on the idea that individuals that behave in a similar fashion will have a similar fitness. Thus, similarity of individuals is taken at the performance level, in contrast to commonly employed approaches which are either based on similarity of genotypes or, less frequently, phenotypes. The approximation performs a weighted average of the fitness values of a number of individuals, attaching a confidence level which is based on similarity estimation. The latter is the second contribution of this work, namely a method for estimating the similarity between individuals. This involves carrying out a number of tests consisting of playing a ‘static’ version of the game (with fixed inputs) with the individuals whose similarity is under evaluation and comparing the results. Because the tests involve a limited version of the game, the computational expense of the similarity estimation plus that of the fitness approximation is much lower than that of directly calculating the fitness. The success of the fitness approximation by similarity estimation method for bot evolution in UT2K4 allows us to expect similar results in environments that share the same characteristics.

Pool vs. Island Based Evolutionary Algorithms: An Initial Exploration

This paper explores the scalability and performance of pool and island based evolutionary algorithms, both of them using as a mean of interaction an object store, we call this family of algorithms SofEA. This object store allows the different clients to interact asynchronously, the point of the creation of this framework is to build a system for spontaneous and voluntary distributed evolutionary computation. The fact that each client is autonomous leads to a complex behavior that will be examined in the work, so that the design can be validated, rules of thumb can be extracted, and the limits of scalability can be found. In this paper we advance the design of an asynchronous, fault-tolerant and scalable distributed evolutionary algorithm based on the object store CouchDB. We test experimentally the different options and show the trade-offs that pool and island-based solutions offer.

Empirical Validation of a Gossiping Communication Mechanism for Parallel EAs

The development of Peer-to-Peer (P2P) systems is still a challenge due to the huge number of factors involved. Validation of these systems must be defined in terms of describing the adequacy of the P2P model to the actual environment. This paper focuses on the validation of the Distributed Resource Machine (DRM) as a computational P2P system when applied to Evolutionary Algorithms (EAs ) using exclusively gossip-based mechanisms for communication. The adequacy will be measured by the range in which performance speedup actually takes place. Validation has been carried out by running an empirical performance study based on benchmarking techniques. It shows that it scales only up to a limited and small number of nodes, which is problem-dependent. Furthermore, due to the reason found for this lack of scalability, it seems unlikely that massive scalability takes place.

Prune and Plant: A New Bloat Control Method for Genetic Programming

This paper reports a comparison of several bloat control methods and also evaluates a new proposal for limiting the size of the individuals: a genetic operator called prune and plant. The aim of this work is to prove the adequacy of this new method. Since a preliminary study of the method has already shown promising results, we have performed a thorough study in a set of benchmark problems aiming at demonstrating the utility of the new approach. Prune and plant has obtained results that maintain the quality of the final solutions in terms of fitness while achieving a substantial reduction of the mean tree size in all four problem domains considered. In addition, in one of these problem domains prune and plant has demonstrated to be better in terms of fitness, size reduction and time consumption than any of the other bloat control techniques under comparison.

Strong Typing, Variable Reduction and Bloat Control for Solving the Bankruptcy Prediction Problem Using Genetic Programming

In this chapter we present the application of a genetic programming (GP) algorithm to the problem of bankruptcy prediction. To carry out the research we have used a database that includes extensive information (not only economic) from the companies. In order to handle the different data types we have used Strongly Typed GP and variable reduction. Also, bloat control has been implemented to obtain comprehensible classification models. For comparison purposes we have solved the same problem using a support vector machine (SVM). GP has achieved very satisfactory results, improving those obtained with the SVM.

NodEO, a multi-paradigm distributed evolutionary algorithm platform in JavaScript

After more than fifteen years, JavaScript has finally risen as a popular language for implementing all kind of applications, from server-based to rich internet applications. The fact that it is implemented in the browser and in server-side tools makes it interesting for designing evolutionary algorithm frameworks that encompass both tiers, but besides, they allow a change in paradigm that goes beyond the canonical evolutionary algorithm. In this paper we will experiment with different architectures, client-server and peer to peer to assess which ones offer most advantages in terms of performance, scalability and ease of use for the computer scientist. All implementations have been released as open source, and besides showing that the concept of working with evolutionary algorithms in JavaScript can be done efficiently, we prove that a master-slave parallel architecture offers the best combination of time and algorithmic improvements in a parallel evolutionary algorithm that leverages JavaScript implementation features.

Modeling Pheromone Dispensers Using Genetic Programming

Mating disruption is an agricultural technique that intends to substitute the use of insecticides for pest control. This technique consists of the diffusion of large amounts of sexual pheromone, so that the males are confused and mating is disrupted. Pheromones are released using devices called dispensers. The speed of release is, generally, a function of time and atmospheric conditions such as temperature and humidity. One of the objectives in the design of the dispensers is to minimise the effect of atmospheric conditions in the performance of the dispenser. With this objective, the Centro de Ecologia Quimica Agricola (CEQA) has designed an experimental dispenser that aims to compete with the dispensers already in the market. The hypothesis we want to validate (and which is based on experimental results) is that the performance of the CEQA dispenser is independent of the atmospheric conditions, as opposed to the most widely used commercial dispenser, Isomate CPlus. This was done using a genetic programming (GP) algorithm. GP evolved functions able to describe the performance of both dispensers and that support the initial hypothesis.