Heuristic constraints enforcement for training of and rule extraction from a fuzzy/neural architecture. II. Implementation and application

Abstract

This paper is the second of two companion papers. The foundations of the proposed method of heuristic constraint enforcement on membership functions for knowledge extraction from a fuzzy/neural architecture was given in Part I. Part II develops methods for forming constraint sets using the constraints and techniques for finding acceptable solutions that conform to all available a priori information. Moreover, methods of integration of enforcement methods into the training of the fuzzy-neural architecture are discussed. The proposed technique is illustrated on a fuzzy-AND classification problem and a motor fault detection problem. The results indicate that heuristic constraint enforcement on membership functions leads to extraction of heuristically acceptable membership functions in the input and output spaces. Although the method is described on a specific fuzzy/neural architecture, it is applicable to any realization of a fuzzy inference system, including adaptive and/or static fuzzy inference systems.

Design of fuzzy rule-based classifiers with semantic cointension

“…To face this problem the common practice is to rely on considerations about the basic structure of the involved fuzzy rule-based models. This leads to the formulation of a number of interpretability constraints (see [33] for a survey), and many approaches have been proposed for interpretability-driven design of fuzzy models based on such constraints [16,17,6,18,36,47].…”

Interpretability of Fuzzy Systems: Current Research Trends and Prospects

“…Industry. Industrial applications could take advantage from interpretable fuzzy systems when there is the need of explaining the behavior of complex systems and phenomena, like in fault detection [11]. Also, control plans for systems and processes can be designed with the aid of fuzzy systems.…”

Interpretability constraints for fuzzy information granulation

“…If -as often occurs -the adaptation process is iterative, the intermediate models emerging after each iteration may not satisfy the interpretability constraints. This is usually the case when adaptation is carried out by means of regularized learning [166] or when interpretability constraints are applied after several adaptation stages for efficiency pursuits [167]. However, when the adaptation process can be halted at every stage (e.g.…”