Genetic Algorithms with Heuristic - Knight's Tour Problem

Genetic algorithms are good search techniques for finding solutions to NP problems. However, their high degree of randomness sometimes fails to guide the search towards finding solutions within reasonable costs. The Knight’s tour problem is an example of where pure GAs fail (practically) to find solutions. Combining GAs with other approaches on the other hand can highly improve their efficiency. In this paper, we present a new approach for improving the performance and effectiveness of GAs by applying heuristic. In our method, while the population is evolving using GA operators, a heuristic is applied to evolve the population either to full solutions or to better partial solutions to be used in generating the next population. The performance of this approach is compared against standard depth-first search with backtracking, heuristic alone, repair alone and also with genetic algorithms with repair. We applied our method to the Knight’s tour problem using binary encoding. Results showed that our method generated more solutions compared to other approaches, and that solutions were found early in the search stages with less probability of failing to find a solution.