Postfix Hardware Evaluation Unit for Genetic Algorithms: Application in Fuzzy Clustering

Genetic algorithms are a class of stochastic optimization techniques inspired by biological evolution processes. The power of GAs for solving complex problems is highly used in the design of parallel problem solving machines. High parallelism needs higher number of parallel processors to be used simultaneously. This approach may be costly in terms of efficiency and utilization of processors. GAs are time costly processes mainly because of their time consuming evaluation operations. Development of a low cost hardware evaluation unit may help reducing time complexities of GAs. In this paper, an attempt is made to show how the fitness evaluation operation of any genetically encoded problem can be performed by using a simple hardware. Our hardware uses a postfix notation of the fitness expression. Since, in GAs, the same function is evaluated for a fairly large number of times, we need to compile a postfix expression only once at the beginning of the genetic optimization process. We performed some simulation experiments on function optimization problems, in general. As an example of combinatorial optimization, we considered the fuzzy clustering problem.