一种求解旅行商问题的LV回复式神经网络模型

A Lotka-Volterra Recurrent Neural Network Model for Solving TSP

旅行商问题是一个组合优化问题。首先,构造一个能量函数来表示旅行商问题,该能量函数的能量最小点对应一条有效的近似最优访问路径;然后,构造一种LV神经网络模型来求解该能量函数的能量最小点。实验结果表明,本文提出的LV神经网络模型能够收敛到能量最小点,并且与Hopfield网络相比,该LV神经网络模型具有更好的求解性能。

The traveling salesman problem （TSP） is a combinational optimization problem. Firstly, constructing an energy func- tion to express the TSP, and a valid near optimization traveling path of TSP could be obtained at an energy minimum point of the energy function. After that, a Lotka-Volterra （LV） recurrent neural network （RNN） model is proposed to solve energy minimum points of the energy function. Experiments show that the proposed LV RNN model should converge to the energy minimum points of the corresponding energy function, and that compared with Hopfield network, the proposed LV RNN has better performance on solving TSP.