Keke Chen
Associate Professor at CSEE
Department of Computer Science and Electrical Engineering
University of Maryland, Baltimore County
Office: ITE 360
Phone: TBD (office)
LinkedIn: https://www.linkedin.com/in/keke-chen-29b5bb2/
Email: kekechen AT umbc . edu
ShortBio: I got my PhD in Computer Science from Georgia Tech and have been working in both industry (Yahoo! Labs) and academia.
I am looking for highly motivated students who are interested in AI, security and privacy, data science, and distributed computing. Our graduates have joined Meta, TikTok, Google, HP, Amazon, IBM, etc.
What's New
Can we evaluate the privacy risk of a data contributor before he/she shares data with a machine learning task? Our demo system "Demo: FT-PrivacyScore: Personalized Privacy Scoring Service for Machine Learning Participation" by Yuechun Gu and Keke Chen, will appear in ACM CCS 2024 in Salt Lake City.
TEE-MR: Developer-Friendly Data Oblivious Programming for Trusted Execution Environments, by Mubashwir Alam and Keke Chen, will appear in one of the oldest security journals: Computers and Security, 2024.
After four wonderful years with Marquette and Milwaukee, we moved to UMBC in August 2024!
Collaboration on small-scale machine learning models on cancer-detection devices: Allison Scarbrough, Keke Chen, Bing Yu, "Designing a use-error robust machine learning model for quantitative analysis of diffuse reflectance spectra", to appear in the Journal of Biomedical Optics, 2024
TEE for confidential graph mining in the cloud: Mubashwir Alam and Keke Chen, "TEE-Graph: efficient privacy and ownership protection for cloud-based graph spectral analysis", to appear in Frontiers in Big Data, 2023
Demos on image disguising methods (CCS23) and TEE-MR (ICDCS23). Check the links
Developing oblivious solutions for Trusted Execution Environments to improve the resilience to side-channel attacks: "Making Your Program Oblivious: a Comparative Study for Side-channel-safe Confidential Computing" to appear in CLOUD 2023.
"GAN-Based Domain Inference Attack" is a new attack designed to enhance model-targeted attacks, such as model inversion attacks and membership inference attacks. It helps identify the most relevant domains for a target model if the attacker has no prior domain knowledge. It will appear in AAAI 2023. Congratulations, Yuechun (Ethan) Gu!
Glad to be a part of the CyberWIN team for the first "NSF CyberCorps Scholarship for Service" award in Wisconsin. The project will start in 1/2023.
Thanks to the NSF CICI program for supporting our project "Confidential Computing in Reproducible Collaborative Workflows" starting in 1/2023!