科研之友
微信
新浪微博
Facebook
分享链接
摘要
A new recurrent neural network is proposed for solving quadratic and linear programming problems, which is derived from two salient existing neural networks. One of the predecessors has lower structural complexity but were not shown to be Capable of solving degenerate QP problems including LP problems while the, other does not have this limitation but has higher structural complexity. The proposed model inherits the merits of both models and thus serves as a competitive alternative for solving QP and LP problems. Numerical simulations are provided to demonstrate Hie performance of the model and validate the theoretical results.
- 出版日期2009
- 单位智能技术与系统国家重点实验室; 清华大学
