An effectiveness analysis of transfer learning for the concept drift problem in malware detection

作者：

Highlights：

• Transfer learning is effective in overcoming concept drift in malware classification.

• Performance improvement varies across models; tree-type models benefit the least.

• Transfer learning also improves models when labeled target domain data is available.

摘要

•Transfer learning is effective in overcoming concept drift in malware classification.•Performance improvement varies across models; tree-type models benefit the least.•Transfer learning also improves models when labeled target domain data is available.

论文关键词：Transfer learning,Machine learning,Malware detection,Concept drift,Cybersecurity

论文评审过程：Received 15 July 2021, Revised 28 August 2022, Accepted 28 August 2022, Available online 3 September 2022, Version of Record 6 September 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118724