Solving the Assignment Problem with the Improved Dual Neural Network

Xiaolin Hu, Jun Wang

Abstract: During the last two decades, several neural networks have been proposed for solving the assignment problem, and most of them either consist of O(n 2) neurons (processing units) or contain some time varying parameters. In the paper, based on the improved dual neural network proposed recently, we present a new assignment network with 2n neurons and some constant parameters only. Compared with the existing neural networks for solving the assignment problem, its more favorable for implementation. Numerical simulation results indicate that the time complexity of the network is O(n).