The research paper introduces MORPH, a pseudo-label-based concept drift adaptation method designed for neural networks to mitigate the challenges in malware detection. Concept drift refers to the shift in the underlying distribution of the testing dataset that deviates from the training dataset. This is a significant issue in malware detection as new malware variants emerge or adversaries attempt to evade detection methods. The proposed method, MORPH, reduces annotation efforts and improves over existing works in automated concept drift adaptation for malware detection, demonstrated through extensive experimental analysis of Android and Windows malware datasets.
Publication date: 23 Jan 2024
Project Page: https://arxiv.org/abs/2401.12790v1
Paper: https://arxiv.org/pdf/2401.12790
