Security and Privacy in the Era of Machine Learning

Security and Privacy in the Era of Machine Learning

Xiaoyong (Brian) Yuan

Machine learning is playing an increasingly important role in our daily lives, serving a slew of novel applications for smart homes, office, and healthcare. However, these applications are known to be vulnerable to various attacks, including traditional cybersecurity attacks and unique threats for breaking the integrity and confidentiality of machine learning systems. Such threats hinder the widespread adoption of machine learning techniques. In this talk, I will present our recent efforts to explore and mitigate two critical machine learning threats: model stealing attacks and membership inference attacks. First, with a high business value, machine learning models have become essential components for various commercialized machine learning services, such as Machine Learning as a Service (MLaaS). We identified a data-agnostic model stealing attack that extracts a functionally equivalent copy from the machine learning services and compromises the confidentiality and integrity of machine learning models. Second, I will delve into membership inference attacks, which reveal the confidential information of training data. We performed the first analysis of membership inference attacks on neural network pruning, an essential technique for reducing large-size machine learning models' computation and memory requirements. The talk will conclude by discussing potential research directions for security and privacy in the era of machine learning

Dr. Xiaoyong (Brian) Yuan is an assistant professor at the College of Computing at Michigan Technological University. Dr. Yuan received his PhD in Computer Science from the University of Florida in 2020, his master's degree in Software Engineering from Peking University in 2015, and BS in Mathematics from Fudan University in 2012. His research spans the fields of machine learning, security and privacy, and Internet of Things. His research has been funded by multiple NSF awards as a PI. He is the recipient of the ORAU Ralph E. Powe Junior Faculty Enhancement Award 2022 and the Michigan Tech ICC Achievement Award 2022. He is currently serving as an associate editor for the IEEE Transactions on Neural Networks and Learning Systems (TNNLS).