Professor Patrick McDaniel
Computer Science and Engineering
Pennsylvania State University
More info here
Title: Attacks, Defenses, and Impacts of Machine Learning in Adversarial Settings
Advances in machine learning have enabled to new applications and services to process inputs in previously unthinkably complex environments. Autonomous cars, data analytics, adaptive communication systems and self-aware software systems are now revolutionizing markets and blurring the lines between computer systems and real intelligence. In this talk, I consider evolving use of machine learning in security-sensitive contexts and explore why many systems are vulnerable to nonobvious and potentially dangerous manipulation. Here, we examine sensitivity in any application whose misuse might lead to harm—for instance, forcing adaptive network in an unstable state, crashing an autonomous vehicle or bypassing an adult content filter. I explore the use of machine learning in this area particularly in light of discoveries in the creation of adversarial samples and defenses against them, and posit on future attacks on machine learning. The talk is concluded with a discussion of the unavoidable vulnerabilities of systems built on probabilistic machine learning, and outline areas for offensive and defensive research in the future.
Patrick McDaniel is a Distinguished Professor in the School of Electrical Engineering and Computer Science and Director of the Institute for Networking and Security Research at the Pennsylvania State University. Professor McDaniel is a Fellow of the IEEE and ACM and serves as the program manager and lead scientist for the Army Research Laboratory's Cyber-Security Collaborative Research Alliance. Patrick’s research centrally focuses on a wide range of topics in security and technical public policy. Prior to joining Penn State in 2004, he was a senior research staff member at AT&T Labs-Research.
Professor Ninghui Li
Professor of Computer Science
More information here
Title: Differential Privacy: Potential and Limitations
Differential privacy (DP) has been increasingly accepted as the de facto standard for data privacy in the research community. Recently, techniques for satisfying DP in the local setting have been deployed by Google and Apple. In this setting, each user perturbs her date before sending it out. This enables the gathering of statistics while preserving privacy of every user, without relying on trust in a single data curator. In the talk, we present our optimized protocols for satisfying DP in the local setting, and discuss the power and limitations of such techniques. Furthermore, we will explore whether satisfying DP is indeed sufficient for protecting privacy, in both the standard setting and the local setting.
Ninghui Li is a Professor of Computer Science at Purdue University. His research interests are in security and privacy. Prof. Li is Chair of ACM Special Interest Group on Security, Audit and Control (SIGSAC), and is serving on the editorial boards of ACM Transactions on Privacy and Security (TOPS), Journal of Computer Security (JCS), and ACM Transactions on Internet Technology. He has also served as Program Chair for 2014 and 2015 ACM Conference on Computer and Communications Security (CCS), ACM's flagship conference in the field of security and privacy.