Menu Close

V. S. Subrahmanian

Securing Online Platforms with Dynamic Adversary-Aware Techniques

Most past work on securing online platforms (e.g. Amazon, Twitter) have been based on classification models. Such classifiers try to predict things such as whether a product review is fraudulent, whether an account is malicious, whether a post contains hate speech, and more. However, they are static. They do not assume that adversaries will evolve continuously. In this talk, I will provide case studies of two applications of how adversaries might evolve continuously in an effort to “beat” such classifiers. Our first application will focus on how malicious entities engaging in selling fraudulent reviews on platforms such as Amazin, while our second application will study how nation states might run influence operations on social platforms. In both cases, we provide defense strategies to mitigate the effect of such malicious operations.

V. S. Subrahmanian

is the Walter P. Murphy Professor of Computer Science at the McCormick School of Engineering, Northwestern University and Buffett Faculty Fellow at the Northwestern Roberta Buffett Institute for Global Affairs. He is also the head of the Northwestern Security and AI Laboratory (NSAIL). Prior to this, he was The Dartmouth College Distinguished Professor in Cybersecurity, Technology, and Society at Dartmouth College, and tenured Professor in the University of Maryland’s Computer Science Department, and Director of the University of Maryland’s Institute for Advanced Computer Studies. His work stands at\\ the intersection of data-driven AI for increased security, policy, and business needs. His keynote speech will be on AI and security problems.