Brian Carse

Evolving fuzzy rule based controllers using genetic algorithms

Abstract The synthesis of genetics-based machine learning and fuzzy logic is beginning to show promise as a potent tool in solving complex control problems in multi-variate non-linear systems. In this paper an overview of current research applying the genetic algorithm to fuzzy rule based control is presented. A novel approach to genetics-based machine learning of fuzzy controllers, called a Pittsburgh Fuzzy Classifier System # 1 (P-FCS1) is proposed. P-FCS1 is based on the Pittsburgh model of learning classifier systems and employs variable length rule-sets and simultaneously evolves fuzzy set membership functions and relations. A new crossover operator which respects the functional linkage between fuzzy rules with overlapping input fuzzy set membership functions is introduced. Experimental results using P-FCS 1 are reported and compared with other published results. Application of P-FCS1 to a distributed control problem (dynamic routing in computer networks) is also described and experimental results are presented.

Fuzzy-XCS: A Michigan Genetic Fuzzy System

The issue of finding fuzzy models with an interpretability as good as possible without decreasing the accuracy is one of the main research topics on genetic fuzzy systems. When they are used to perform online reinforcement learning by means of Michigan-style fuzzy rule systems, this issue becomes even more difficult. Indeed, rule generalization (description of state-action relationships with rules as compact as possible) has received a great attention in the nonfuzzy evolutionary learning field (e.g., XCS is the subject of extensive ongoing research). However, the same issue does not appear to have received a similar level of attention in the case of Michigan-style fuzzy rule systems. This may be due to the difficulty in extending the discrete-valued system operation to the continuous case. The intention of this contribution is to propose an approach to properly develop a fuzzy XCS system for single-step reinforcement problems.

Fast Evolutionary Learning of Minimal Radial Basis Function Neural Networks Using a Genetic Algorithm

A hybrid algorithm for determining Radial Basis Function (RBF) networks is proposed. Evolutionary learning is applied to the non-linear problem of determining RBF network architecture (number of hidden layer nodes, basis function centres and widths) in conjunction with supervised gradient-based learning for tuning connection weights. A direct encoding of RBF hidden layer node basis function centres and widths is employed. The genetic operators utilised are adapted from those used in recent work on evolution of fuzzy inference systems. A parsimonious allocation of training sets and training epochs to evaluation of candidate networks during evolution is proposed in order to accelerate the learning process.

A Fuzzy Classifier System Using the Pittsburgh Approach

This paper describes a fuzzy classifier system using the Pittsburgh model. In this model genetic operations and fitness assignment apply to complete rule-sets, rather than to individual rules, thus overcoming the problem of conflicting individual and collective interests of classifiers. The fuzzy classifier system presented here dynamically adjusts both membership functions and fuzzy relations. A modified crossover operator for particular use in Pittsburgh-style fuzzy classifier systems, with variable length rule-sets, is introduced and evaluated. Experimental results of the new system, which appear encouraging, are presented and discussed.

Special issue on “Genetic Fuzzy Systems: Recent Developments and Future Directions”

Fuzzy systems have demonstrated their ability to solve different kinds of problems in various application domains. In most cases, the key for success was their ability to incorporate human expert knowledge. In the 1990s, despite the previous palmy history, a certain interest for the study of fuzzy systems with added learning capabilities emerged. Two of the most important approaches have been the hybridization attempts made in the framework of soft computing, where different techniques such as neural networks and evolutionary computation provide fuzzy systems with learning ability. A Genetic Fuzzy System (GFS) is basically a fuzzy system, usually a fuzzy rule-based system (FRBS), augmented by a learning process based on evolutionary computation, which includes Genetic Algorithms, Genetic Programming, or Evolutionary Strategies, among others. Evolutionary learning processes cover different levels of complexity according to the structural changes produced in the fuzzy system by the algorithm. The fusion of these population-based, robust search algorithms with a representation that offers linguistic interpretability such as fuzzy systems provides a powerful paradigm for computational intelligence research. Among the main reasons to bet the use of evolutionary algorithms instead other optimization/learning techniques to design fuzzy systems we can highlight the following. First, they provide a powerful and flexible search capability (such as the use of multiple objectives, constrained objectives, or multimodal objectives) that allow them to address a wide range of problems. Second, they can process flexible representation structures (such as mixed coding schemes or constrained representation) that allow them to deal with almost any kind of fuzzy system. Besides, they can run on distributed and cellular architectures, perform incremental learning, and easily hybrid with other techniques for complex tasks. The field of GFS has now reached a stage of maturity after the earliest papers were published 17 years ago. Although the maturity of the GFS field means it is now being applied to an ever growing number of real-world applications, there are many basic issues yet to be resolved and there is an active and vibrant worldwide community of researchers working on these issues.

Application of multi-agent technology to fault diagnosis of power distribution systems

When a fault occurs in a power system, the protective relays detect the fault and trip appropriate circuit breakers, which isolate the affected equipment from the rest of the power system. Fault diagnosis of power systems is the process of identifying faulty components and/or sections by analysing observable symptoms (telemetry messages). As the domain is characterised by dynamic situations, extensive telemetering, complex operations, and distribution of lines and substations over a large geographical area, it is difficult to tackle fault diagnosis problems through the strength and capability of a single intelligent system. This paper describes an experimental multi-agent system developed for and aimed at a computer-supported fault diagnosis in electricity distribution networks. The system is based on a hierarchy of five agents that cooperate with each other to diagnose a fault. A set of detailed case studies is presented, and the results obtained suggest that an agent-based approach is very efficient and has a good potential for real-time application.

Evolving radial basis function neural networks using a genetic algorithm

Most research to date using genetic algorithms to evolve neural networks has focused on the multi-layer perceptron. Altemative neural network approaches such as the radial basis function network, and their representations appear to have received relatively little attention as grist for the GA mill. This is perhaps surprising since, for example, the radial basis function network has also been proved to be universal function approximator. Here we focus on evolution of radial basis function networks. While the multilayer perceptron network approximates functions through global interaction between network nodes, the radial basis function network uses local interactions between network nodes. It is suggested, that this difference may be of significance in tem of epistatic interactions in encoded genomes for the two types of network, which affects the ability of the genetic algorithm to evolve successful networks. A representation and attendant genetic operators for evolving radial basis function networks are proposed, drawing on recent work on evolutionary fuzzy logic systems. Experimental results in applying a hybrid leaming technique, using a genetic algorithm for evolving the radial basis function hidden layer (number of hidden nodes and hidden node centres and widths) and supervised leaming for tuning of network connection weights, are presented.

CURRENT AND FUTURE TRENDS IN FEATURE SELECTION AND EXTRACTION FOR CLASSIFICATION PROBLEMS

In this article, we describe some of the important currently used methods for solving classification problems, focusing on feature selection and extraction as parts of the overall classification task. We then go on to discuss likely future directions for research in this area, in the context of the other articles from this special issue. We propose that the next major step is the elaboration of a theory of how the methods of selection and extraction interact during the classification process for particular problem domains, along with any learning that may be part of the algorithms. Preferably this theory should be tested on a set of well-established benchmark challenge problems. Using this theory, we will be better able to identify the specific combinations that will achieve best classification performance for new tasks.

Fuzzy Q-Learning with an adaptive representation

Reinforcement learning (RL) is learning how to map states to actions so as to maximise a numeric reward signal. Fuzzy Q-learning (FQL) extends the RL technique Q-learning to large or continuous problems and has been applied to a wide range of applications from data mining to robot control. Typically, FQL uses a uniform or pre-defined internal representation provided by the human designer. A uniform representation usually provides poor generalisation for control applications, and a pre-defined representation requires the designer to have an in-depth knowledge of the desired control policy. In this paper, the approach taken is to reduce the reliance on a human designer by adapting the internal representation, to improve the generalisation over the control policy, during the learning process. A hierarchical fuzzy rule based system (HFRBS) is used to improve the generalisation of the control policy through iterative refinement of an initial coarse representation on a classical RL problem called the mountain car problem. The process of adapting the representation is shown to significantly reduce the time taken to learn a suitable control policy.

An approach to the reduction of contact bounce using fuzzy control

When contacts close, any contactor bounce is undesirable since this causes arcing which shortens the life of the contacts. A method of reducing contactor bounce using fuzzy control is proposed. A model of a widely used contactor has been created for experimental purposes. This model is described. The proposed fuzzy controller takes as input the relative position and relative velocity of the contacts and provides as output a voltage which controls the rate of closing of the contacts. Experimental results are presented for the simulation of contact bounce with and without fuzzy control. It is demonstrated that, even with a relatively simple fuzzy rule base, bounce duration and number of bounces can be effectively reduced.