Discrete optimizations using graph convolutional networks

In this paper we discuss the use of graph deep learning in solving quadratic assignment problems (QAP). The quadratic assignment problem is an NP hard optimization problem. We shall analyze an approach using Graph Convolutional Networks (GCN). We prove that a specially designed GCN produces the optimal solution for a broad class of assignment problems. By appropriate training, the class of problems correctly solved is thus enlarged. Numerical examples compare this method with other simpler methods.