|August 23, 2023
|Download as iCal:
Malicious mobile applications steal user credentials, mine cryptocurrency on user devices, and lock user data to demand ransom.
Detecting such malware is challenging even for the most secure application stores, as evident from the recent incidents of malware penetrating the official Android store. In this talk, I will first map the challenges faced by program-analysis-based and machine-learning-based techniques when identifying malicious applications. These include (a) the inability of analysis-based approaches to reason about dynamically-loaded, obfuscated, and hard-to-trigger code and (b) limited reliability and explainability of machine-learning-based approaches. I will then highlight opportunities to advance the state-of-the-art in malware detection and describe our current work on (a) developing an efficient, just-in-time, on-device dynamic malware detection approach and (b) understanding the effects of training data on the reliability of machine-learning-based detection.
Finally, I will put this work in context of the broader effort of my research group to improve quality, reliability, and security of software and AI-based systems.
Julia Rubin is an Associate Professor at the Department of Electrical and Computer Engineering at the University of British Columbia, Canada and a Canada Research Chair in Trustworthy Software. She leads the UBC Research Excellence Cluster on Trustworthy ML. Julia received her PhD in Computer Science from the University of Toronto and worked as a postdoctoral researcher in the Computer Science and Artificial Intelligence Lab at MIT. She also spent almost 10 years in industry, working for IBM Research, where she was a research staff member and a research group manager. Julia's research interests are in software quality, software security, and reliability of software and AI systems.
Together with her research group, she develops solutions that enable construction of reliable software systems in an efficient manner.
Currently, her work focuses on security and integrity of mobile and cloud-based systems, as well as on robustness, explainability, and fairness of AI systems. Her work in these areas won six Distinguished/Best Paper Awards at major conferences, including ASE, ISSTA, ICST, and ICSME.