Académique Documents
Professionnel Documents
Culture Documents
Within the evolutionary computation field, a new 2. For each subspace S,, obtain a set of candi-
paradigm has been recently proposed, coevolu- date consequents (i.e., linguistic terms of the
tionary algorithms [7]. They involve two or more output variable) BSto build the correspond-
species (populations) that permanently interact ing linguistic rule.
among them by a coupled fitness. Thereby, in
spite of each species has its own coding scheme In this contribution, the set of candidate con-
and reproduction operators, when an individual sequents for S, is defined as follows:
must be evaluated, its goodness will be calculated
BS= {Bk E I gels E EL where
considering some individuals of the other species.
This coevolution makes easier to find solutions to vB1 E PB,(Y'"> 2 PB~(Y")).
complex problems.
3. Perform a combinatorial search among these
Intuitively, we may decompose the problem of de-
sets looking for the combination of conse-
riving a proper KB for an FRBS into two sub-
quents (one for each subspace) with the best
tasks: to obtain linguistic rule symbolic represen-
global accuracy.
tations (learning the FU3) and to define member-
s h i ~function shapes (learning the DB). There-
fore, our algorithm consists of two species that On the other hand, the derivation of the DB in-
cooperate to build the whole solution. volves determining the shape of each membership
function. These shapes will have a high influ-
In the following subsections, a formulation for
ence in the FRBS performance. In this contri-
both learning tasks and the cooperative coeve
bution, we will consider triangular-shaped mem-
lutionary algorithm are introduced.
bership functions.