The first thing that needs to be performed when attempting to
solve a task by means of cooperative coevolution is the division
of the problem into several components that can be separately
handled. When a potential solution for each component is
evaluated, it is connected to those of the other complementary
components for the purpose of building a complete solution.
Similarly, in the end of the algorithm, complete results of the
task are formed by gathering solutions for each of the components.
A complete solution for a classification problem may consist of a
set of rules. Consequently, in the novel approach of cooperative
coevolutionary classification, each population evolves one type of
rule. Experiments are driven on three data sets and obtained
results demonstrate the promise of proposed technique.