一种基于粗糙集理论的粗糙神经网络构造方法

提出在BP神经网络中使用粗糙集理论网络的设计，由于粗糙集理论有强大的数值分析能力,而BP神经网络具有准确的逼近收敛能力和较高的精度，所以通过两者的结合，可以得到一种可理解性好，计算简单，收敛速度快的神经网络模型，这种神经网络的算法的主要过程为：首先利用粗糙集能力去发现给定数据集的一些规则，然后根据这些规则构造神经网络稳含层的神经元个数，最后用BP算法迭代求了网络的各种参数，完成网络的设计，本文最后给出了一个三维非线性函数的实例进一步验证了网络的正确性。

An Approach to Construct a Rough Neural Networks Based on Rough Set Theory

A new scheme of knowledge encoding in a BP neural networks using rough set theoretical concepts is described. Rough set theory has a powerful capability of qualitative analysis, while BP neural networks possess the capability with the good approaching convergence and the higher accuracy. By combining those advantages of the two theories. A kind of neural networks with good understandability, simple computation and exact accuracy can be constructed. The key process of the learning algorithm for the rough neural networks is as follows: find a set of rules from the given training data by rough set theory, construct the neurons in the hidden layers according to those rules, and then learn the arguments of the neural networks with BP algorithm. Finally, a satisfying example is also presented.