Data Science Seminar on Oct 30th

Title:  Blockchain-Empowered Secure Decentralized Machine Learning in a Heterogeneous Environment

Speaker: Dr. Qianlong Wang

 

Abstract: In the big data era, federated learning has become one of the most critical applications, allowing different parties to collaborate to obtain better learning models without sharing their own data. However, there are several main concerns about the traditional federated learning systems. First, most existing systems are distributed and need a central server to coordinate learning. Such a central server may raise security concerns, e.g., a single point of failure, etc. Second, federated learning relies on several assumptions/requirements, e.g., independent and identically distributed (i.i.d.) data and model homogeneity. Since more and more edge devices can train lightweight models with local data, such models are usually heterogeneous. In this talk, I will present a blockchain-empowered federated learning framework that enables learning in a fully decentralized manner while considering model and data heterogeneity. In particular, a federated learning framework with a heterogeneous calibration process, i.e., Model and Feature Calibration (FL-MFC), is introduced to enable collaboration among heterogeneous models. I will further present a two-level mining process using blockchain to allow decentralized learning to defend against critical system attacks, e.g., Byzantine attacks, while achieving effective learning performance.

 

Bio: Dr. Qianlong Wang is currently an Assistant Professor in the Department of Electrical and Computer Engineering at Old Dominion University, Norfolk, VA. Before this, he was an Assistant Professor in the Department of Computer and Information Sciences at Towson University, Towson, MD, from 2021 to 2024. He received the B.E. degree in Electrical Engineering from Wuhan University, China, in 2013, the M.E. degree in Electrical and Computer Engineering from Stevens Institute of Technology, Hoboken, NJ, in 2015, and the Ph.D. degree in Computer Engineering from Case Western Reserve University, Cleveland, OH, in 2021. His research interests include robust and secure deep learning networks, security and privacy in distributed and decentralized systems, e.g., Cyber-Physical Systems (CPS) and Internet of Things (IoT), and related applications, e.g., smart transportation, smart city, and smart healthcare. He is a recipient of the NSF CRII Award in 2023.