A generalized neural network for solving minimax problems with nonsmooth cost functions

This paper investigates a class of minimax problems, in which the cost functions are nonsmooth. A generalized neural network for solving the minimax problems was proposed, and its convergence was proven based on the nonsmooth analysis. The rate of convergence was discussed by virtue of the łojasiewicz inequality. Two numerical examples were given to illustrate the efficiency of the theoretical results.