Logic-based genetic programming for multi-modal solutions

We present a new method of logic-based genetic programming (LBGP) using Prolog programming framework. Using the intrinsic mechanism of backtracking in Prolog, we are able to (i) design flexible genetic operations effective to the smaller sizes of populations, and (ii) to maintain the population diversity of individual programs with both active structures (or exons) and inactive structures (or introns) generated by the genetic operations. We apply the method to obtaining multiple solutions for multi-modal problems. We validate the effectiveness of LBGP to get multiple solutions on tree constructional and tile world problems.