粗糙集模糊神经网络软件质量预测

Rough set based fuzzy neural network algorithm for software quality prediction

以面向对象的软件度量为研究对象,首先采用SOM神经网络离散化度量元因子矩阵数据,接着对于得到的离散化的矩阵数据采用粗糙集理论的属性约简算法进行属性约简,然后根据约简得到规则构造模糊神经网络的网络结构,并采用BP算法对网络进行训练,最后通过仿真实验验证了该算法。

Taking the object oriented software as example ,we apply the SOM neural network to discretize data matrix and then minimize the data with rough set reduction algorithm .Based on the simplifying rules , We build the fuzzy neural network structure trained by BP algorithm . The experiments show that the algorithm is feasible .