lambda-opt neural approaches to quadratic assignment problems

In this article, we propose new analog neural approaches to combinatorial o ptimization problems, in particular, quadratic assignment problems (QAPs). Our proposed methods are based on an analog version of the lambda-opt heuri stics, which simultaneously changes assignments for lambda elements in a pe rmutation. Since we can take a relatively large lambda value, our new metho ds can achieve a middle-range search over possible solutions, and this help s the system neglect shallow local minima and escape from local minima. In experiments, we have applied our methods to relatively large-scale (N = 80- 150) QAPs. Results have shown that our new methods are comparable to the pr esent champion algorithms; for two benchmark problems, they are obtain bett er solutions than the previous champion algorithms.