Cross-Project Software Defect Prediction Based on Feature Selection and Transfer Learning
| Autor: | Jingfeng Xue, Tianwei Lei, Weijie Han |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Computer science
business.industry Process (computing) 020207 software engineering Feature selection 02 engineering and technology computer.software_genre Data set Software Software bug Filter (video) 020204 information systems 0202 electrical engineering electronic engineering information engineering Data mining business Transfer of learning Cross project computer |
| Zdroj: | Machine Learning for Cyber Security ISBN: 9783030624620 ML4CS (3) |
| Popis: | Cross-project software defect prediction solves the problem that traditional defect prediction can’t get enough data, but how to apply the model learned from the data of different mechanisms to the target data set is a new problem. At the same time, there is the problem that information redundancy in the training process leads to low accuracy. Based on the difference of projects, this paper uses MIC to filter features to solve the problem of information redundancy. At the same time, combined with the TrAdaboost algorithm, which is based on the idea of aggravating multiple classification error samples, this paper proposes a cross-project software prediction method based on feature selection and migration learning. Experimental results show that the algorithm proposed in this paper has better experimental results on AUC and F1. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|