基于SA-BP神经网络的软件缺陷预测模型的研究  

Research on Software Defect Prediction Model Based on SA-BP

作　　者：尹然[1,2] 丁晓明[1,2] 李小亮[1,2] 梅莹[1,2] 

机构地区：[1]西南大学计算机与信息科学学院,重庆400715 [2]重庆市软件评测中心有限公司,重庆400715

出　　处：《西南师范大学学报（自然科学版）》2013年第8期147-152,共6页Journal of Southwest China Normal University(Natural Science Edition)

摘　　要：探讨了传统BP神经网络的模型与结构,并针对BP神经网络容易陷入局部最优的缺陷,提出用模拟退火技术代替局部梯度下降法修正网络权值的SA-BP算法,用于构建SA-BP神经网络的软件缺陷预测模型,并通过实验证明了SA-BP神经网络模型应用于软件缺陷预测的有效性.This paper deals with the discussion on the model and structure of traditional BP neural network,and with the analyses of the weakness that BP neural networks is easy to fall into local optimal solution.In order to solve the problem,a software defect prediction model has been put forward based on SA-BP,with simulated Annealing to replace the local gradient descent method.The experiment results indicate that the SA-BP has a higher prediction precision than the traditional software defect prediction model.

关 键 词：SA-BP神经网络 软件缺陷预测模型 模拟退火 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]