Search

In this paper, a hybrid clustering and classification algorithm is obtained by exploring the specific statistical model of a hyperplane classifier. We show how the seamless integration of the clustering component allows a substantial cost decrease in the training stage, without impairing the performance of the classifier. The algorithm is also robust to outliers and deals with training errors in a natural and efficient manner.

A binary classification algorithm, called Probabilistic Vector Machine – PVM, is proposed. It is based on statistical measurements of the training data, providing a robust and lightweight classification model with reliable performance. The proposed model is also shown to provide the optimal binary classifier, in terms of probability of error, under a set of loose conditions regarding the data distribution. We compare PVM against GEPSVM and PSVM and provide evidence of superior performance on a number of datasets in terms of average accuracy and standard deviation of accuracy.

The protein structure prediction problem consists in finding good computational algorithms for prediction of protein native states. This paper applies the Particle Swarm Optimization (PSO) algorithm to predict the tertiary structure of proteins in lattice models. We propose a novel discrete PSO variant designed for lattice-based protein folding models. We present three lattice based models and two folding encodings, which are tested in different combinations on six proteins. The results indicate that the new algorithm performs very efficient and finds very good proteins conformations.

This paper proposes two general techniques for adapting operators in a genetic algorithm: one dynamically adjusts their rates, while the other customizes their specific way of operation. We show how these techniques can be integrated into a single evolutionary system, called Integrated-Adaptive Genetic Algorithm (IAGA). The IAGA exhibits fewer input parameters to adjust than the original GA, while being able to automatically adapt itself to the particularities of the optimization problem it tackles. We present a proof-of-concept implementation of this technique for royal road functions.

This paper describes work in progress to increase the performance of a memetic timetabling system. The features looked at are two directed mutation operators, targeted mutation and a structured population that facilitates parallel implementation. Experimental results are given that show good performance improvements with directed and targeted mutation, and acceptable first results with the structure population.

This paper presents a new genetic representation for timetabling with evolutionary algorithms. The representation involves the use of suggestion lists for the placement of events into timeslots. A set of recombination operators is defined for the new representation, and experimental results are given to compare the performance of the operators with each other and with a system not using suggestion lists.