A. Chong

FUZZY COGNITIVE MAPS WITH GENETIC ALGORITHM FOR GOAL-ORIENTED DECISION SUPPORT

Fuzzy cognitive maps are signed directed graphs used to model the evolution of scenarios with time. FCMs can be useful in decision support for predicting future states given an initial state. Genet...

Improvement of the Cluster Searching Algorithm in Sugeno and Yasukawa's Qualitative Modeling Approach

Fuzzy modeling has become very popular because of its main feature being the ability to assign meaningful linguistic labels to the fuzzy sets in the rule base. This paper examines Sugeno and Yasukawas qualitative modeling approach, and addresses one of the remarks in the original paper. We propose a cluster search algorithm that can be used to provide a better projection of the output space to the input space. This algorithm can efficiently identify two or more fuzzy clusters in the input space that have the same output fuzzy cluster.

On the Fuzzy Cognitive Map attractor distance

Fuzzy cognitive map (FCM) has commonly been used as a prediction tool. The FCM forward chains have been used to find answers to what-if questions. The process starts with the encoding of the what-if question into a stimulus vector. The vector goes through a series of vector-matrix multiplication until the FCM converges to one of the FCM attractors. The attractor is the answer to the initial question. There are several types of FCM attractors. The usefulness of the different types of attractors relies heavily on the users objectives and interpretations. This paper presents the theoretical discussion on distance measurement among the various FCM attractor distances. Subsequently the FCM attractor distance (FCMAD) based on genetic algorithm is proposed. The use of this distance in FCM goal oriented analysis and FCM learning is discussed. Experiment results have confirmed the effectiveness of the proposed technique.

A hybrid approach for solving the cluster validity problem

A hybrid approach for solving the cluster validity problem is proposed. The proposed technique uses a cluster validity index in conjunction with a merging index to find the optimal number of clusters in a set of data. The technique does not place any restriction on the fuzzy clustering technique and the cluster validity index used. This paper examines the use of fuzzy c-means clustering and the validity index proposed by Fukuyama and Sugeno (1989). Experiments are carried out to verify the effectiveness of the proposed technique. It is shown in this paper that the proposed technique is more reliable than Fukuyama and Sugenos original validity index.

A histogram-based rule extraction technique for fuzzy systems

We propose a histogram-based rule extraction technique using straightforward histogram-based clustering that produces trapezoidal clusters that are well suited for the rule extraction purpose. Two experiments were carried out to validate the feasibility and effectiveness of the proposed technique and show that the rule base generated by the proposed technique is reasonably accurate.

Hierarchical fuzzy classifier for bioinformatics data

In this research, a preliminary study of the application of hierarchical fuzzy rule-based classifier for protein secondary structure prediction has been carried out. The use of a hierarchical structured rulebase alleviates, to some extent, the problem of rule explosions that has prevented the use of traditional fuzzy system in many biomedical related problems. As part of the study, a hierarchical fuzzy classifier was built from a set of training data. Although the accuracy of the classifier is far from comparable to the current established techniques, the experiment has successfully confirmed the feasibility of the application of the hierarchical classifier for protein structure prediction. This calls for further research to further improve the accuracy of the rule-based classifier. The advantages of using the rule-based classifier as compared to other artificial intelligent techniques for protein structure prediction are also discussed in the paper.

Extending the Decision Accuracy of a Bioinformatics System

We introduce a simple fuzzy technique to improve the prediction decision accuracy of a bioinformatics neural network system from the literature for protein structure prediction. We also describe an unsound assumption made by the authors of the neural network system, and propose a fuzzy hybrid solution, which eliminates the need for this assumption and can further enhance performance.

Histogram-Based Fuzzy Clustering and Its Comparison to Fuzzy C-Means Clustering in One-Dimensional Data

In this paper, a histogram fuzzy clustering technique (HFC) has been proposed. The technique is designed specifically for one-dimensional data clustering. HFC is composed of two main components: trapezoidal cluster approximation, and bin width determination. By conducting experiments on two sets of real-world petroleum data, the effectiveness of HFC and the Fuzzy c-Means Clustering technique has been compared. It is found that the performance of HFC and FCMC varies across different sets of data. In some data, HFC can perform significantly better than FCMC. It is concluded that despite its limitation to clustering only one-dimensional data, HFC can be very useful in the rules extraction problem domain, especially in clustering the output space, which can be assumed to be always one-dimensional. Several other advantages of HFC, namely the technique is straightforward and computational efficient, are also discussed in the paper

Subspace clustering for hierarchical fuzzy system construction

Hierarchical fuzzy systems are proposed to deal with the rule explosion problem of traditional fuzzy systems. The inference operations of the fuzzy systems are well established. The next step is to tackle the problem of finding subspaces for automated hierarchical fuzzy system construction. We propose a clustering technique designed specifically for this purpose. It is both theoretically and experimentally confirmed that the algorithm has reasonable accuracy and scalability.