Catálogo de publicaciones - libros

In brain-computer interfaces (BCIs), a feature selection approach using an adaptive genetic algorithm (AGA) is described in this paper. In the AGA, each individual among the population has its own crossover probability and mutation probability. The probabilities of crossover and mutation are varied depending on the fitness values of the individuals. The adaptive probabilities of crossover and mutation are propitious to maintain diversity in the population and sustain the convergence capacity of the genetic algorithms (GAs). The performance of the AGA is compared with those of the Standard GA (SGA) and the Filter method in selecting feature subset for BCIs. The results show that the classification accuracy obtained by the AGA is significantly higher than those obtained by other methods. Furthermore, the AGA has a higher convergence rate than the SGA.

This paper describes an approach to online simplification of evolved programs in genetic programming (GP). Rather than manually simplifying genetic programs after evolution for interpretation purposes only, this approach automatically simplifies programs during evolution. In this approach, algebraic simplification rules, algebraic equivalence and prime techniques are used to simplify genetic programs. The simplification based GP system is examined and compared to a standard GP system on a regression problem and a classification problem. The results suggest that, at certain frequencies or proportions, this system can not only achieve superior performance to the standard system on these problems, but also significantly reduce the sizes of evolved programs.

In this paper we propose a hybrid model which includes both first principles differential equations and a least squares support vector machine (LS-SVM). It is used to forecast and control an environmental process. This inclusion of the first principles knowledge in this hybrid model is shown to improve substantially the stability of the model predictions in spite of the unmeasurability of some of the key parameters. Proposed hybrid model is compared with both a hybrid neural network(HNN) as well as hybrid neural network with extended kalman filter(HNN-EKF). From experimental results, proposed hybrid model shown to be far superior when used for extrapolation compared to HNN and HNN-EKF.

In this paper, we propose a novel method for rapid speaker adaptation called speaker support vector selection (SSVS). By taking gaussian mixture model (GMM) as speaker model, the speakers acoustically close to the test speaker are selected .Different from other selection method, just computing the likelihood between models, we utilizing support vector machines (SVM) to obtain a ‘more optimal speaker subset’. Such selection is dynamically determined according to the distribution of reference speakers close the test. Furthermore, a single-pass re-estimation procedure conditioned on the selected speakers is shown. This adaptation strategy was evaluated in a large vocabulary speech recognition task. The presented method improves the relative accuracy rates by 13% compared to the baseline system.

Many researchers focus on enhancing the flexibility of the workflow management system. This paper shows how simulation plays an important role in improving the flexibility of temporal logic based workflow specification (TLWS) model. Detecting the TLWS model’s demand for changing and deciding what changes to carry out are very difficult. Simulation analysis can help to do these. After a task is finished, its time, cost and quality will be computed. And each will be compared with dual threshold values. If the value is below the bottom threshold, an exception will be thrown. If it is above this threshold, but below the top threshold, a warning will be sent out. If the value is above the top threshold, the execution is excellent. Workflow specification documents also need to be translated into simulation specification documents by using XLST. The utility of using simulation in improving the flexibility of TLWS model is outstanding.

This paper presents a novel method for predictive lossless image compression via evolving a set of switches, which can be implemented easily by intrinsic evolvable hardware mode. A set of compounded mutations for binary chromosome through combining the local asexually reproducing with multiple mean step size search was proposed, and a gradually approach method for evolving larger scale images was fabricated. Experimental results show that the proposed method can reduce the computing time much more, and can scale up the image size increasing up to 70 times with relative slower increase speed of computing time.

In a 3D intelligent virtual environment, multi 3D virtual characters interact with emotion and construct a 3D artificial society. Modeling emotion interaction is a challenging topic for virtual artificial society. Nonverbal emotion interaction is a direct communication manner for virtual characters. A cognitive model of virtual character is presented. A 3D virtual character has a cognitive architecture with built-in knowledge that control emotion and the response to outer stimuli. Some new concepts on nonverbal social interaction are set up.

This paper explores the feasibility of computer simulation of evolving populations of social animals in nature, both from the anatomical and socio-environmental viewpoints, addressing the gap between the algorithms for evolution of digital objects, and the evolution of species in nature. The main components of ant body are mathematically described within the function representation framework; the parameters directly determine both the visual characteristics of the ant as well as the body characteristics encoded by the genome. The environmental diversification of ant subspecies is studied for fungus-growing ants, in which single-queen mating reproduction couples with large size of accessory male glands, while multiple-queen mating correlates to large size of accessory testes. Our results show that within an environment of restricted resources, both competing modes of sexual reproduction survive. The frequency with which either mode becomes dominant in the population is driven by the value of the mutation probability. The function representation model should be useful also in the simulation of other simple animal species, because of the ease in relating the genome parameters to computer visualization tools.

The evolutionary dynamics of minority games based on three generic networks have been investigated : Kauffman’s NK networks (Kauffman nets), growing directed networks (s), and growing directed networks with a small fraction of link reversals (s). We show that the dynamics and the associated phase structure of the game depend crucially on the structure of the underlying network. The dynamics on s is very stable for all values of the connection number , in contrast to the dynamics on Kauffman’s NK networks, which becomes chaotic when >=2. The dynamics of s, on the other hand, is near critical. Under a simple evolutionary scheme, the network system with a “near” critical dynamics evolves to a high level of global coordination among its agents; this suggests that criticality leads to the best performance. For Kauffman nets with >3, the evolutionary scheme has no effect on the dynamics (it remains chaotic) and the performance of the MG resembles that of a random choice game (RCG).

In this piece of research, genetic algorithms are put forward for solving the HIP game. The proposed parallel approach manipulates candidate solutions via selection and mutation; no crossover has been employed. The population is limited to one candidate solution per generation, thus keeping the computational complexity of the approach to a minimum. It is shown that the proposed approach is superior to the approaches reported in the literature: solutions are more speedily provided while the frequency of finding a solution is significantly higher.