Speaker
Description
As Android continues to dominate the global mobile market, cybercriminals
increasingly target its vast user base with sophisticated malware. In this
presentation, we propose an interpretable framework for Android malware
detection that leverages language model to analyze a range of
features—including app manifests, API calls, and opcode sequences. By
integrating feature analysis techniques, our approach not only achieves high
detection accuracy but also provides critical insights into which features
drive classification decisions. We will share empirical results
demonstrating the method’s effectiveness on real-world datasets, discuss the
benefits of interpretability for security practitioners, and explore how
these findings can inform the next generation of mobile threat defense
systems.
Optional: Speaker / convener biography
Hantang Zhang is a doctoral student in Software Engineering and Security at
Umeå university. His research interests include natural language processing
(NLP) and software security. Currently, he focuses on leveraging
transformer-based language models and feature analysis frameworks to enhance
the accuracy and transparency of Android malware detection.
