Catálogo de publicaciones - libros

In this paper we study the differences in performance among different implementations, in Java, of the data structures used in an evolutionary algorithm (EA). Typical studies on EAs performance deal only with different abstract representations of the data and pay no attention to the fact that each data representation has to be implemented in a concrete programming language which, in general, offers several possibilities, with differences in time consumed, that may be worthy of consideration.

The paper presents a method that allows an intelligent multiagent system to coordinate and negotiate their actions in order to achieve a common goal. Each individual agent consists of several autonomous components that allow the agent to perceive and react to its environment, to plan and execute an action, and to negotiate with other agents in an intelligent manner. A heuristic approach for conflict solution is presented, which is used for coordination of a society of independently acting agents in common environment. The application of the method is shown on an intelligent multiagent robotic system realizing complex transfer operations simultanously.

Location-aware applications, such as indoor robot navigation or human activity monitoring, require the location estimation of moving elements, by using ultrasonic, infrared or radio signals received from sensors deployed in the workplace. These sensors are commonly arranged in regular lattices on the ceiling. However, this configuration is not optimal for location estimation using trilateration techniques, in terms of positioning precision, maximum coverage and minimum singular cases. This paper shows how non-regular optimal sensor deployments, generated with a new meta-heuristic optimization methodology (Diversified Local Search - DLS), outperforms regular lattices for trilateration.

Virtual sensors are a key element in many modern control and diagnosis systems, and their importance is continuously increasing; if there are no appropriate models available, virtual sensor design has to be based on data. Structure identification using Genetic Programming is a method whose ability to produce models of high quality has been shown in many theoretical contributions as well as empirical test reports. One of its most prominent shortcomings is relatively high runtime consumption; additionally, one often has to deal with problems such as overfitting and the selection of optimal models out of a pool of potential models that are able to reproduce the given training data.

In this article we present a sliding window approach that is applicable for Genetic Programming based structure identification; the selection pressure, a value measuring how hard it is to produce better models on the basis of the current population, is used for triggering the sliding window behavior. Furthermore, we demonstrate how this mechanism is able to reduce runtime consumption as well as to help finding even better models with respect to test data not considered by the training algorithm.

The Rural Postman Problem consists of finding a shortest tour containing all edges in a subset, in the subgraph induced by some subset of nodes. In general, the problem is NP-hard, since the Traveling Salesman Problem can be easily transformed into it. The Traveling Salesman Problem consist of finding a shortest closed tour which visits all the cities in a given set. Artificial Neural Networks have been applied to solve the Traveling Salesman Problem in the last years. In this work we propose to apply self-organizing feature maps to solve the first problem, transforming it previously into the second.

Variable population sizing techniques are rarely considered in the theory of Genetic Algorithms. This paper discusses a new variant of adaptive population sizing for this class of Evolutionary Algorithms. The basic idea is to adapt the actual population size depending on the actual ease or difficulty of the algorithm in its ultimate goal to generate new child chromosomes that outperform their parents.

In this paper the author presents three approaches to parallel Tabu Search, applied to several instances of the Capacitated Vehicle Routing Problem with soft Time Windows (CVRPsTW). The Tabu Search algorithms are of two kinds: Two of them are parallel with respect to functional decomposition and one approach is a collaborative multisearch TS. The implementation builds upon a framework called Distributed metaheuristics or DEME for short. Tests were performed on an SGI Origin 3800 supercomputer at the Johannes Kepler University of Linz, Austria.

The balance of exploration and exploitation is the essence of any successful meta-heuristic. The represents a simple form of this general dilemma. This paper describes two heuristic optimization methods that use a simple yet efficient allocation strategy for the bandit problem called to control the optimization process.

The algorithms are applied to the well known and the results compared to the results of other successful meta-heuristics for this scheduling problem.

Signal coding and pulse compression techniques have been recently introduced in Local Positioning Systems as a means to enhance the measurement precision of these systems and to increase their operation frequency. This work presents a Genetic Algorithm that performs the search for an optimal family of binary codes, to be used in a system based on ultrasonic technology. The developed algorithm takes into account the transduction effect of the emitters on the correlation properties of the modulated family to obtain a set of codes that exhibits a superior performance than other families previously used.

RND (Radio Network Design) is an important problem in mobile telecommunications (for example in mobile/cellular telephony), being also relevant in the rising area of sensor networks. This problem consists in covering a certain geographical area by using the smallest number of radio antennas achieving the biggest cover rate. To date, several radio antenna models have been used: square coverage antennas, omnidirectional antennas that cover a circular area, etc. In this work we use omnidirectional antennas. On the other hand, RND is an NP-hard problem; therefore its solution by means of evolutionary algorithms is appropriate. In this work we study different evolutionary approaches to tackle this problem. PBIL (Population-Based Incremental Learning) is based on genetic algorithms and competitive learning (typical in neural networks). DE (Differential Evolution) is a very simple population-based stochastic function minimizer used in a wide range of optimization problems, including multi-objective optimization. SA (Simulated Annealing) is a classic trajectory descent optimization technique. Finally, CHC is a particular class of evolutionary algorithm which does not use mutation and relies instead on incest prevention and disruptive crossover. Due to the complexity of such a large analysis including so many techniques, we have used not only sequential algorithms, but also grid computing with BOINC in order to execute thousands of experiments in only several days using around 100 computers.