Ernesto Costa

An Empirical Comparison of Particle Swarm and Predator Prey Optimisation

In this paper we present and discuss the results of experimentally comparing the performance of several variants of the standard swarm particle optimiser and a new approach to swarm based optimisation. The new algorithm, which we call predator prey optimiser, combines the ideas of particle swarm optimisation with a predator prey inspired strategy, which is used to maintain diversity in the swarm and preventing premature convergence to local suboptima. This algorithm and the most common variants of the particle swarm optimisers are tested in a set of multimodal functions commonly used as benchmark optimisation problems in evolutionary computation.

GVR: A New Genetic Representation for the Vehicle Routing Problem

In this paper we analyse a new evolutionary approach to the vehicle routing problem. We present Genetic Vehicle Representation (GVR), a two-level representational scheme designed to deal in an effective way with all the information that candidate solutions must encode. Experimental results show that this method is both effective and robust, allowing the discovery of new best solutions for some well-known benchmarks.

An Immune System-Based Genetic Algorithm to Deal with Dynamic Environments: Diversity and Memory

The standard Genetic Algorithm has several limitations when dealing with dynamic environments. The most harmful limitation as to do with the tendency for the large majority of the members of a population to convergence prematurely to a particular region of the search space, making thus difficult for the GA to find other solutions when changes in the environment occur. Several approaches have been tested to overcome this limitation by introducing diversity in the population or through the incorporation of memory in order to help the algorithm when situations of the past can be observed in future situations. In this paper, we propose a GA inspired in the immune system ideas in order to deal with dynamic environments. This algorithm combines the two aspects mentioned above: diversity and memory and we will show that our algorithm is also more adaptable and accurate than the other algorithms proposed in the literature.

Multidimensional Knapsack Problem: A Fitness Landscape Analysis

Fitness landscape analysis techniques are used to better understand the influence of genetic representations and associated variation operators when solving a combinatorial optimization problem. Five representations are investigated for the multidimensional knapsack problem. Common mutation operators, such as bit-flip mutation, are employed to generate fitness landscapes. Measures such as fitness distance correlation and autocorrelation are applied to examine the landscapes associated with the tested genetic encodings. Furthermore, additional experiments are made to observe the effects of adding heuristics and local optimization to the representations. Encodings with a strong heuristic bias are more efficient, and the addition of local optimization techniques further enhances their performance.

Evolutionary Algorithms for Dynamic Environments: Prediction Using Linear Regression and Markov Chains

In this work we investigate the use of prediction mechanisms in Evolutionary Algorithms for dynamic environments. These mechanisms, linear regression and Markov chains, are used to estimate the generation when a change in the environment will occur, and also to predict to which state (or states) the environment may change, respectively. Different types of environmental changes were studied. A memory-based evolutionary algorithm empowered by these two techniques was successfully applied to several instances of the dynamic bit matching problem.

New Directions for Intelligent Tutoring Systems

1 Foundations.- New Perspectives on Cognition and Instructional Technology.- A Genetic Structure for the Interaction Space.- COLAPSES: A Modular Architecture and Language for Modelling Meta-Control and Uncertainty.- Computational Mathetics: the Missing Link in Intelligent Tutoring Systems Research?.- Whats in an ITS? A Functional Decomposition.- 2 Student Modelling.- Meta-Reasoning and Student Modelling.- Machine Learning, Explanation-Based Learning and Intelligent Tutoring Systems.- The Central Importance of Student Modelling to Intelligent Tutoring.- 3 ITS: Principles and Practices.- Student Models, Scratch-Pads, and Simulation.- A Framework for Instructional Planning and Discourse Modelling in Intelligent Tutoring Systems.- Uses of ITS: Which Role for the Teacher?.- 4 Belief Systems.- A Belief Revision Model of Repair Sequences in Dialogue.- A Structure for Epistemic States.- Building an Intelligent Second Language Tutoring System from Whatever Bits You Happen to Have Lying Around.- 5 Interaction Among Agents.- Negotiating Goals in Intelligent Tutoring Dialogues.- Integration of Knowledge in Multi-Agent Environments.- Facing Hard Problems in Multi-Agent Interactions.- List of Contributors.

On the influence of GVR in vehicle routing

A comparative study is made between a new evolutionary approach for the Vehicle Routing Problem (VRP) and a standard evolutionary model, based on Path Representation. Genetic Vehicle Representation (GVR) is the new two-level representational scheme designed to deal in an effective way with all the information needed by candidate solutions. Experimental results, obtained from a set of VRP instances, show performance improvements when GVR is used.

An Evolutionary Approach to the Zero/One Knapsack Problem: Testing Ideas from Biology

The transposition mechanism, widely studied in previous publications, showed that when used instead of the standard crossover operators, allows the genetic algorithm to achieve better solutions. Nevertheless, all the studies made concerning this mechanism always focused the domain of function optimization. In this paper, we present an empirical study that compares the performances of the transposition-based GA and the classical GA solving the 0/1 knapsack problem. The obtained results show that, just like in the domain of function optimization, transposition is always superior to crossover.

On the Evolution of Evolutionary Algorithms

In this paper we discuss the evolution of several components of a traditional Evolutionary Algorithm, such as genotype to phenotype mappings and genetic operators, presenting a formalized description of how this can be attained. We then focus on the evolution of mapping functions, for which we present experimental results achieved with a meta-evolutionary scheme.

Improving prediction in evolutionary algorithms for dynamic environments

The addition of prediction mechanisms in Evolutionary Algorithms (EAs) applied to dynamic environments is essential in order to anticipate the changes in the landscape and maximize its adaptability. In previous work, a combination of a linear regression predictor and a Markov chain model was used to enable the EA to estimate when next change will occur and to predict the direction of the change. Knowing when and how the change will occur, the anticipation of the change was made introducing useful information before it happens. In this paper we introduce mechanisms to dynamically adjust the linear predictor in order to achieve higher adaptability and robustness. We also extend previous studies introducing nonlinear change periods in order to evaluate the predictors accuracy.