Proposta de aplicação das redes neurais artificiais paraconsistentes como classificador de sinais utilizando aproximação funcional

Abstract

This work has the objective of developping methods for applications of Paraconsistents Artificiais Ncurals Networks - PANNs. Firstly, it will be introduced some topies about Paraconsistent Logic (with prominence for the functional part of the cells that compose it) and also about PANNs (with prominence for the Paraconsistents Artificiais Neurals Cells). Later, it is made an analysis about the conduct of the Apprenticeship Paraconsistent Artificial Neural Cell, when it is inserted in its entrance pattems that represents discrcet signal'values. From this results, it develops for the cntirely reproduetion of a signal through functional approach. Thcn, it is formed the Paraconsistents Artificiais Neurals Units, with three cells, capables to learn a pattem ( signal ) and also capable to compare these patterns with others introduced on the unit. The unit outlets are manipulated according to its application prccision. The group of these units form the Paraconsistents Artificiais Neurals Networks for Signals Classifiers, capable to learn and to store various types of pattems (signals) on the network, and also capable to classify these patterns as pertaining or not the same network when introduced pattems on it. The perspective is that the results presented and the used mcthodology attcnd on the divulgation for others types applications with PANNs.