DeepTrust: Multi-Step Classification through Dissimilar Adversarial Representations for Robust Android Malware Detection
This addresses the critical problem of adversarial robustness for Android malware detection systems, offering a novel approach that significantly improves security without sacrificing accuracy on clean data.
The paper tackles the vulnerability of machine learning models to adversarial examples in Android malware detection by introducing DeepTrust, a metaheuristic that sequences classifiers to maximize representation divergence, achieving state-of-the-art results with up to 266% better performance under attacks and a false positive rate below 1%.
Over the last decade, machine learning has been extensively applied to identify malicious Android applications. However, such approaches remain vulnerable against adversarial examples, i.e., examples that are subtly manipulated to fool a machine learning model into making incorrect predictions. This research presents DeepTrust, a novel metaheuristic that arranges flexible classifiers, like deep neural networks, into an ordered sequence where the final decision is made by a single internal model based on conditions activated in cascade. In the Robust Android Malware Detection competition at the 2025 IEEE Conference SaTML, DeepTrust secured the first place and achieved state-of-the-art results, outperforming the next-best competitor by up to 266% under feature-space evasion attacks. This is accomplished while maintaining the highest detection rate on non-adversarial malware and a false positive rate below 1%. The method's efficacy stems from maximizing the divergence of the learned representations among the internal models. By using classifiers inducing fundamentally dissimilar embeddings of the data, the decision space becomes unpredictable for an attacker. This frustrates the iterative perturbation process inherent to evasion attacks, enhancing system robustness without compromising accuracy on clean examples.