Fuzzy Rule Base Generation Using Discretization of Membership Functions and Neural Network

Autor:	Vytautas Pilkauskas, Tadas Kraujalis, Henrikas Pranevičius, Germanas Budnikas
Rok vydání:	2014
Předmět:	Fuzzy classification Fuzzy rule business.industry Computer science Fuzzy set Pattern recognition computer.software_genre Fuzzy logic Defuzzification Fuzzy number Artificial intelligence Data mining Decision table business computer Membership function
Zdroj:	Communications in Computer and Information Science ISBN: 9783319119571 ICIST
DOI:	10.1007/978-3-319-11958-8_13
Popis:	Paper presents a technique for fuzzy rule extraction. It applies a division of a feature space into fuzzy grids and a selection of discrete values as inputs for neural network. A neural network generates fuzzy rules that are simplified using a decision table analysis tool Prologa. The tool detects and fixes cases of redundancy and ambivalence in a fuzzy rule base. A case study contains an illustration of the proposed technique and a comparison of the results to other sources. A comparative analysis of a productivity of traffic light controllers developed using an expert rule base and a rule base formed using our technique is given. Iris classification problem is considered too. Comparison results prove better accuracy of the technique suggested.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_________::44803f03cb5e7638e2021f30bc4b5b7a https://doi.org/10.1007/978-3-319-11958-8_13 Zobrazit plný text záznamu