A. J. Yuste

Knowledge Acquisition in Fuzzy-Rule-Based Systems With Particle-Swarm Optimization

Knowledge acquisition is a long-standing problem in fuzzy-rule-based systems. In spite of the existence of several approaches, much effort is still required to increase the efficiency of the learning process. This study introduces a new method for the fuzzy-rule evolution that forms an expert system knowledge: the knowledge acquisition with a swarm-intelligence approach (KASIA). Specifically, this strategy is based on the use of particle-swarm optimization (PSO) to obtain the antecedents, consequences, and connectives of the rules. To test the feasibility of the suggested method, the inverted-pendulum problem is studied, and results are compared for two of the most extensively used methodologies in machine learning: the genetic-based Pittsburgh approach and the Q-learning-based strategy, i.e., state-action-reward-state-action (SARSA). Moreover, KASIA is analyzed as a learning strategy in fuzzy-rule-based metascheduler design for grid computing, and performance is compared with other scheduling strategies based on genetic learning and existing scheduling approaches, i.e., EASY-backfilling and ESG+local periodical search. To be more precise, simulation results prove the fact that the proposed strategy outperforms classical learning approaches in terms of final results and computational effort. Furthermore, the main advantage is the capability to control convergence and its simplicity.

A fuzzy rule-based meta-scheduler with evolutionary learning for grid computing

Grid computing is increasingly emerging as a promising platform for large-scale problems solving in science, engineering and technology. Nevertheless, a major effort is still required to harness the high potential performance of such computational framework and in this sense, an important challenge is to develop new strategies that efficiently address scheduling on the distributed, heterogeneous and shared environment of grids. Fuzzy rule-based systems (FRBSs) models are dynamic and are currently attracting the interest of scheduling research community to obtain near-optimal solutions on grids. However, FRBSs performance is strongly related to the quality of their knowledge bases and thus, with the knowledge acquisition process. Due to the inherent dynamic nature and the typical complex search spaces of grids, automatically finding a high-quality knowledge base that accurately describes the fuzzy system is extremely relevant. In this work, we propose a scheduling system for grids considering a novel learning strategy inspired by Michigan and Pittsburgh approaches that applies genetic algorithms (GAs) to evolve the fuzzy rule bases and improves the classical learning strategies in terms of computational effort and convergence behaviour. In addition, experimental results show that the proposed schema significantly outperforms other extensively used scheduling strategies.

Genetic fuzzy rule-based scheduling system for grid computing in virtual organizations

One of the most challenging problems when facing the implementation of computational grids is the system resources effective management commonly referred as to grid scheduling. A rule-based scheduling system is presented here to schedule computationally intensive Bag-of-Tasks applications on grids for virtual organizations. There exist diverse techniques to develop rule-base scheduling systems. In this work, we suggest the joining of a gathering and sorting criteria for tasks and a fuzzy scheduling strategy. Moreover, in order to allow the system to learn and thus to improve its performance, two different off-line optimization procedures based on Michigan and Pittsburgh approaches are incorporated to apply Genetic Algorithms to the fuzzy scheduler rules. A complex objective function considering users differentiation is followed as a performance metric. It not only provides the conducted system evaluation process a comparison with other classical approaches in terms of accuracy and convergence behaviour characterization, but it also analyzes the variation of a wide set of evolution parameters in the learning process to achieve the best performance.

Genetic Fuzzy Rule-Based meta-scheduler for Grid computing

The growing interest in grids technologies for the solving of large-scale computational problems leads related framework improvement. One of the challenging problems in Grid computing is the efficient resources utilization and allocation of tasks, i.e. scheduling problem. Fuzzy Rule-Based Systems (FRBSs) have recently proved to be a competitive alternative for the development of scheduling systems, outperforming extensively used scheduling strategies such as EASY Backfilling or Greedy. However, FRBSs-based schedulers performance strongly depends on their data bases quality and a major effort is still required for the knowledge acquisition process improvement. This paper presents a fuzzy rule-based meta-scheduler incorporating a new genetic approach for the learning process. Concretely, the suggested learning strategy is inspired by classical rule evolution strategies, Pittsburgh and Michigan approaches. Experimental results show that further accuracy in the learning process of fuzzy meta-schedulers can be achieved without significantly increasing the associated computational effort.

Learning of Fuzzy Rule-Based Meta-schedulers for Grid Computing with Differential Evolution

Grid computing has arisen as the next-generation infrastructure for high demand computational applications founded on the collaboration and coordination of a large set of distributed resources. The need to satisfy both users and network administrators QoS demands in such highly changing environments requires the consideration of adaptive scheduling strategies dealing with inherent dynamism and uncertainty. In this paper, a meta-scheduler based on Fuzzy Rule-Based Systems is proposed for scheduling in grid computing. Moreover, a new learning strategy inspired by stochastic optimization algorithm Differential Evolution (DE), is incorporated for the evolution of expert system knowledge or rules bases. Simulation results show that knowledge acquisition process is improved in terms of convergence behaviour and final result in comparison to other evolutionary strategy, genetic Pittsburgh approach. Also, the fuzzy meta-scheduler performance is compared to other extended scheduling strategy, EASY-Backfilling in diverse criteria such as flowtime, tardiness and machine usage.

A Method to Minimize Distributed PSO Algorithm Execution Time in Grid Computer Environment

This paper introduces a method to minimize distributed PSO algorithm execution time in a grid computer environment, based on a reduction in the information interchanged among the demes involved in the process of finding the best global fitness solution. Demes usually interchange the best global fitness solution they found at each iteration. Instead of this, we propose to interchange information only after an specified number of iterations are concluded. By applying this technique, it is possible to get a very significant execution time decrease without any loss of solution quality.