Automatic analysis of signals with symbolic content

This paper presents a set of methods for helping in the analysis of signals with particular features that admit a symbolic description. The methodolog y is based on a general discrete model for a symbolic processing subsystem, which is fuzzyfied by means of a fuzzy inference system. In this framework a number of design problems have been approached. The curse of dimensional ity problem and the specification of adequate membership functions are the main ones. In addition, other strategies, which make the design process sim pler and more robust, are introduced. Their goals are automating the produc tion of the rule base of the fuzzy system and composing complex systems fro m simpler subsystems under symbolic constrains. These techniques are applie d to the analysis of wakefulness episodes in the sleep EEG. In order to sol ve the practical difficulty of finding remarkable situations from the outpu ts of the symbolic subsystems an unsupervised adaptive learning technique ( FART network) has been applied. (C) 2000 Elsevier Science B.V. All rights r eserved.