Delayed chaotic neural network with annealing controlling for maximum clique problem

In this paper, we propose a delayed chaotic neural network with annealing controlling strategies (DCNN-AC) to solve the NP-complete maximum clique problem (MCP). We point out some flaws in the variable delayed neural network proposed by Chen, and demonstrate that DCNN-AC is a powerful chaotic neural network through analyzing its single neural model and its “beautiful” chaotic dynamics. DCNN-AC has richer and more flexible chaotic dynamics and flexible annealing controlling strategies, so that it can be expected to have higher searching ability for globally optimal or near-optimal solutions. The DCNN-AC performance has been verified by simulations on some MCP benchmark instances. The comparisons with some famous proximate algorithms show the superiority of DCNN-AC in terms of the solution quality and the comparable computation time.