文档介绍：Discovering Fuzzy Classification Rules with ic
Programming and Co-Evolution
Roberto R. F. Mendes Fabricio de B. Voznika Alex A. Freitas Julio C. Nievola
PUC-PR
PPGIA - CCET
Av. Imaculada Conceição, 1155
Curitiba - PR, 80215-901 Brazil
{alex, nievola}***@
.br/~alex
+55 41 330-1669
Abstract. In essence, data mining consists of extracting knowledge from
data. This paper proposes a co-evolutionary system for discovering fuzzy
classification rules. The system uses two evolutionary algorithms: a ic
programming (GP) algorithm evolving a population of fuzzy rule sets and a
simple evolutionary algorithm evolving a population of membership function
definitions. The two populations co-evolve, so that the final result of the co-
evolutionary process is a fuzzy rule set and a set of membership function
definitions which are well adapted to each other. In addition, our system also
has some innovative ideas with respect to the encoding of GP individuals rep-
resenting rule sets. The basic idea is that our individual encoding scheme in-
corporates several syntactical restrictions that facilitate the handling of rule
sets in disjunctive normal form. We have also adapted GP operators to better
work with the proposed individual encoding scheme.
1 Introduction
In the context of machine learning and data mining, one popular way of expressing
knowledge consists of IF-THEN rules. This is due to the fact that they are intuitively
comprehensible to a human being [5]. In addition, they represent independent units
of knowledge, so that alterations can easily take place in their contents. IF-THEN
rules posed of two parts. The first part (ponent, or rule antecedent)
corresponds to a conjunction of conditions that, if verified true, imply that the con-
dition contained in the second part (ponent, or rule consequent) is also
considered true.
Rules in their classic format are appropriate when their conditions are constituted
by disc