【摘要】 针对复杂优化问题,研究并提出一种基于深度学习的层次结构与Dropout技术的变参数并行玻尔兹曼算法模型。该算法模型能够有效抑制局域最优问题,在TSP问题求解中获得了较好优化效果,并在经典实例eil51搜索到比迄今已知最优解更优的TSP路径,在pr1002等大规模TSP问题求解中较快搜索到迄今已知最优路径。更多还原

【Abstract】 Aiming at the complex optimization problems,a parallel Boltzmann algorithm based on deep learning hierarchy and Dropout technology is studied and proposed in this paper. This algorithm can effectively restrain the local optimum,obtain a better optimization effect in solving TSP problem,as well as search the optimal TSP path better than hitherto known path in classic examples of eil51 and the optimal path so far known in large-scale TSP problems such as in the instance of pr1002.更多还原