A Hybrid Hopfield Network-Genetic Algorithm Approach for the Lights-up Puzzle

摘要

This paper presents a hybrid genetic algorithm for solving a logic-type puzzle known as lights-up puzzle. The algorithm uses a binary Hopfield neural network to solve part of the puzzle constraints. The genetic algorithm manages other set of constraint as it looks for good quality solution in term of the puzzle's objective function. We show the good performance of our approach in a number of lights-up puzzles instances downloaded from the Internet.