Catálogo de publicaciones - libros

In the near future human beings will be living in a ubiquitous world where all objects such as electronic appliances are networked to each other and robots will provide us with services of every variety by any device through any network, at any place anytime. However, current robots have a number of constraints to become wholly ubiquitous. This research attempts to eliminate the spatial limitations by introducing virtual robots into the physical world. Ubiquitous robot, Ubibot, is introduced to integrate three forms of robots: software robot (Sobot), embeded robot (Embot) and mobile robot (Mobot). Sobot is a virtual robot, which has the ability to move to any place or connect to any device through a network, in order to overcome the spatial limitation. It has the capability to interpret the context and thus interact with the user. Embot is embedded within the environment or in the Mobot. An Embot can recognize the locations of and authenticate the user or robot, and synthesise sensing information. Mobot provides integrated mobile services. The services that will be provided by Ubibot will be seamless, calm and context-aware. This talk will address the basic concepts of Ubibot. A Sobot, called Rity, will be introduced in order to investigate the usability of the proposed concepts. Rity is a 3D synthetic character which exists in the virtual world, has a unique IP address and interacts with human beings through an Embot implemented by a face recognition system using a USB camera.

We propose a fuzzy logic based punctual kriging technique for enhancing images corrupted by Gaussian noise. Punctual kriging is used to generate kernel weights employing the semivariances in the neighborhood of a pixel and empirically determined global semi-variogram. Semivariance is a measure of the degree of spatial differences between samples (pixel values). Superiority of kriging over other methods for noise cancellation in 1-D signals has been established. A quantitative analysis of the kriging technique, for image enhancement as compared to the Wiener filter shows that kriging performs inferior to Wiener filtering for image enhancement. We have proposed a new fuzzy logic based method which substantially improves the performance of the kriging for image enhancement. Experimental results are presented to illustrate the improvement in the results and the effectiveness of the new technique.

In this paper a method for optical quality control of coated steel sheets is proposed. A frequently used test for examining coating quality is the impact resistance test which results in a rapid deformation of a steel sheet sample. The quality of the coating with respect to adhesion can then be rated from an image of the deformed region by examination of the texture on the surface of the steel sheet. In the approach described here the texture analysis is carried out using the fuzzy grey scale correlogram which is a modified version of the colour correlogram developed in the context of content-based image retrieval. The correlogram data provides the basis for a classification using a fuzzy rule base. Experimental results indicate that the fuzzy grey scale correlogram is applicable to the problem discussed in this paper.

As a response to the progress in digital data acquisition and storage technology, along with the limited human capabilities in analyzing and exploiting large amounts of data, the field of knowledge discovery in databases (KDD) has recently emerged as a new research discipline, lying at the intersection of statistics, machine learning, data management, and other areas. According to a widely accepted definition, KDD refers to the non-trivial process of identifying valid, novel, potentially useful, and ultimately understandable structure in data. The central step within the overall KDD process is data mining , the application of computational techniques to the task of finding patterns and models in data. Still, KDD also involves further important steps, notably data preparation, data cleaning, incorporation of prior knowledge, and interpretation of data mining results.

This paper defends the employment of Evolutive Algorithms (EAs) in action games by showing their virtues for both offline and online opponent controlling. The paper proposes (and also compares) several EAs applied offline in the solving of a classical path finding problem and used to provide intelligence to autonomous agents (e.g., the opponents) in an action computer game. The paper also presents an EA that has been successfully employed in real time (i.e., online) in an action game in which a player controls a military vehicle in a hostile enemy region.

When designing artificial neural network (ANN) it is important to optimise the network architecture and the learning coefficients of the training algorithm, as well as the time the network training phase takes, since this is the more time-consuming phase. In this paper an approach to cooperative co-evolutionary optimisation of multilayer perceptrons (MLP) is presented. The cooperative co-evolution is performed on the MLP and training set at the same time. Results show that this co-evolutionary model reaches an optimal MLP with generalization error comparable to those presented by other authors but using a smaller training set, co-evolved with the system.

This paper describes a proposal for increasing performances when using Parallel Genetic Algorithms. We apply a new operator that has been recently described within Genetic Programming domain, the plague. By means of a series of experiments on a benchmark problem, we show that computational effort can be reduced when looking for solutions by means of Parallel GAs.

The Protein Structure Prediction (PSP) problem is considered here. This problem is tackled with evolutionary algorithms (EAs) using the HPNX energy model. A classic penalty-based hybrid EAs, and two hybrid EAs are compared in this context. These hybrid EAs are based on the utilization of backtracking-based operators for mutation and recombination. They represent two different approaches for treating a constrained problem as the PSP: repairing infeasible solutions, and maintaining feasibility at all times. Both hybrid EAs perform better than the penalty-based EA, although the difference is more marked in the case of using the repairing procedure.

In the paper the structure of quasi-copulas on discrete scales is studied. A special attention is given to the study of quasi-copulas with diagonal section. The cardinality of the set of quasi-copulas on L _( n ) determined univocally by the corresponding diagonal section is shown to be the Fibonacci number p _n.

We introduce aggregation operators acting on real interval I. We discuss four basic classes of aggregation operators: conjunctive, disjunctive, averaging and mixed. Several examples are given, some construction methods are recalled. In more details we present a new method for construction of averaging aggregation operators and the general ordinal sum method for aggregation operators.