What question did this study set out to answer?

This research aims to assess the effectiveness of different models and techniques for enhancing Android malware detection robustness.

Adversarial collapse to robust recovery: assessing the impact of robustness and adversarial retraining on Android malware detection

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial evaluates robustness in Android malware detection, indicating the need for better security measures.

This research aims to assess the effectiveness of different models and techniques for enhancing Android malware detection robustness.
Developed a benchmarking framework to evaluate various adversarial attack methodologies.
Utilized MalDroid and Drebin datasets for large-scale evaluation and cross-dataset validation.
Evaluated four models—MLP, CNN, XGBoost, and autoencoder-XGBoost—under various attack scenarios.
MLP achieved 97.94% clean accuracy but dropped by up to 20.55% under feature insertion attacks.
XGBoost demonstrated the best robustness-efficiency trade-off against mixed attacks.
Adversarial retraining improved model robustness across all evaluated systems.

Qasim et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark