两步迭代自返馈神经网络排序模型

本文提出了用神经网络有效地解决排序问题的途径和方法。建立了一个自反馈神经网络排序模型，对于ｎ个元素的排序问题，利用该模型可在两个迭代步内完成，排序时间与排序问题的规模无关。该模型的空间复杂度为Ｏ（ｎ＾２）。

SELF FEEDBACK SORTING NEURAL NETWORK MODEL WITH TWO ITERATIONS

Sorting is a fundamental problem of great theoretical and practical importance in computation. The relationship between neural network(NN) and sorting is characterized by the interaction of enormous elements. It is a new way to employ the parallel processing mechanism of NNs for sorting problem solving, and was analyzed in detail in this paper. In our method, a sorting model based on self feedback NNs was established, including the NN's structure, the sorting algorithm, and the initial constraint, etc. According to the method mentioned, an example with only eight data elements considered was given to describe each sorting step. It is noted that only two iterations are needed for any size of sorting issues by means of, for example, O(n 2) neurons, where n is the sorting size. In other words, the sorting time is independent of the sorting size.