论文部分内容阅读
本文提出并行搜索和规划算法，以及实现它们的高阶二维时态－竟争激励神经网络．这种网络还能实现基于传统符号逻辑的许多问题求解算法．本文的方法克服了通常的神经网络求解优化问题的缺陷．同时，也避免了符号逻辑算法的串行性及符号逻辑Ｓｙｓｔｏｌｉｃ结构复杂性等问题．给出了求解隐式图搜索、ＬＣＳ问题、ＴＳＰ问题及０－１背包问题的实例．
In this paper, parallel search and planning algorithms are proposed, and their high-order two-dimensional tense-competitive excitation neural networks are realized. This network also enables many problem solving algorithms based on traditional symbolic logic. The proposed method overcomes the shortcomings of the conventional neural network in solving optimization problems. At the same time, it avoids the serialization of symbolic logic algorithms and the complexity of Systolic structure of symbolic logic. An example of solving implicit graph search, LCS problem, TSP problem and 0-1 knapsack problem is given.