基于非线性集成深度学习的软件模块风险预测

Risk Prediction of Software Module Based on Nonlinear Integrated Deep Learning

利用当前方法预测软件模块风险时没有对软件模块数据进行预处理,导致预测软件模块风险预测精度较低,效果不佳。为解决上述问题,提出基于非线性集成深度学习的软件模块风险预测方法。利用主成分分析法对软件模块数据进行预处理,在不损失有用信息的前提下可降低数据维度。将处理过的样本训练集进行初始化处理,并更新相应的权重向量,利用向量融合弱分类器构成非线性集成深度学习分类器,通过此分类器得出软件模块中是否包含风险,进而实现软件模块风险预测。实验结果表明,所提方法的预测软件模块风险预测精度较高,有效提升了风险预测效果。

When using the current method to predict software module risk, there is no preprocessing of software module data, resulting in low risk prediction accuracy and poor effect. In order to solve the above problems, a software module risk prediction method based on nonlinear integrated deep learning is proposed. Using principal component analysis to preprocess the software module data can reduce the data dimension without damaging the useful information. The processed sample training set was initialized, and the corresponding weight vector was updated. Vector was applied to fuse weak classifiers to form a nonlinear integrated deep learning classifier. Using to the classifier, the risk in the software module was predicted. Eventually, the risk prediction of the software module was realized. The experimental results show that this method has high prediction accuracy and can improve the prediction effect.