A genetic-neuro algorithm for tiling problems with rotation and/or reflection of figures

Document Type: Research Paper



This paper describes an algorithm for tiling with polyminoes that consider rotation and/or reflection of figures in the steps of 90o. First, we review the previous parallel algorithms for tiling problems. Next, we propose a hybrid approach that is based on genetic algorithms (GA) and artificial neural networks (ANN). In this approach, the production of new members in GA and their evaluation are performed by a Hopfield neural network. Finally we compare our method with the previous works, and show that our method can produce global minima for many problems. The algorithm can be used for solving a variety of 2D-packing problems.