UW-Madison Computer Sciences Professor Somesh Jha was honored with a prestigious “Test of Time” award during the 33rd USENIX Security Symposium.
Held in Philadelphia on August 14-16, USENIX 2024 brought together experts from around the world, who highlighted the latest advances in the security and privacy of computer systems and networks.
Jha received the USENIX Security Test of Time Award for his co-authored paper Privacy in Pharmacogenetics: An End-to-End Case Study of Personalized Warfarin Dosing.
“This paper was highly influential, and was essentially the precursor to the enormous body of literature we have today on empirical measurement of privacy in ML models,” said Kamalika Chaudhuri, Professor of Computer Science at University of California—San Diego and Director at Meta FAIR Labs. “It basically introduces attribute inference and was a precursor to membership inference.”
The USENIX Security Test of Time Award recognizes papers that have had a lasting impact on their fields. To qualify, a paper must have been presented at the USENIX Security Symposium at least 10 years ago. Matt Fredrikson of Carnegie-Mellon University presented the paper with his co-authors at the 2014 USENIX Security Symposium in San Diego.
In their paper, Jha and his colleagues initiated the study of privacy in pharmacogenetics, wherein machine learning models are used to guide medical treatments based on a patient’s genotype and background. As part of this work, they highlighted a new type of attack that could be used to learn information about the models’ confidential training data, which poses risks beyond pharmacogenetics.
Jha said that “this paper investigated privacy concerns in pharmacogentics. Ideas, such as attribute inference and model inversion, originated from this paper.”
Congratulations, Professor Jha!
This story is taken in part from this CyLab story written by Michael Cunningham.