基于BP神经网络的计算机软件缺陷预测方法

Prediction Method of Computer Software Defects Based on BP Neural Network

为提升计算机软件缺陷预测方法精准度,提高软件运行安全,本文提出一种基于反向传播(Back Propagation,BP)神经网络的计算机软件缺陷预测方法。基于BP神经网络进行计算机软件预测中的数据预处理,将软件缺陷分为内容缺陷和需求缺陷,结合过采样与欠采样方法实现类的平衡,引用软件缺陷密度公式计算缺陷数量特征,利用训练梯度直推式支持向量机建立计算机软件缺陷预测模型。实验结果证明:设计方法的计算机软件缺陷预测正确率较高,具有一定有效性与精准度。

In order to enhance the accuracy of computer software defect prediction method and improve the safety of software operation,this paper proposes a computer software defect prediction method based on Back Propagation(BP)neural network.Based on BP neural network for data pre-processing in computer software prediction,software defects are divided into content defects and demand defects,combined with oversampling and undersampling methods to achieve class balance,the software defect density formula is cited to calculate the number of defect features,and a computer software defect prediction model is established using a training gradient direct push support vector machine.The experimental results prove that the design method has a high correct rate of computer software defect prediction with certain validity and accuracy.