Timezone: »

 
Poster
Parameters or Privacy: A Provable Tradeoff Between Overparameterization and Membership Inference
Jasper Tan · Blake Mason · Hamid Javadi · Richard Baraniuk

Tue Nov 29 09:00 AM -- 11:00 AM (PST) @ Hall J #335

A surprising phenomenon in modern machine learning is the ability of a highly overparameterized model to generalize well (small error on the test data) even when it is trained to memorize the training data (zero error on the training data). This has led to an arms race towards increasingly overparameterized models (c.f., deep learning). In this paper, we study an underexplored hidden cost of overparameterization: the fact that overparameterized models may be more vulnerable to privacy attacks, in particular the membership inference attack that predicts the (potentially sensitive) examples used to train a model. We significantly extend the relatively few empirical results on this problem by theoretically proving for an overparameterized linear regression model in the Gaussian data setting that membership inference vulnerability increases with the number of parameters. Moreover, a range of empirical studies indicates that more complex, nonlinear models exhibit the same behavior. Finally, we extend our analysis towards ridge-regularized linear regression and show in the Gaussian data setting that increased regularization also increases membership inference vulnerability in the overparameterized regime.

Author Information

Jasper Tan (Rice University)
Jasper Tan

PhD student at Rice University doing research on computational imaging, computer vision, and machine learning. (I am graduating in December 2022 and am looking for full-time positions in either computational imaging, computer vision, or machine learning!)

Blake Mason (Amazon)

Blake Mason is Doctoral Student at the University of Wisconsin-Madison studying Electrical and Computer Engineering under the advisement of Professor Robert Nowak. Prior to his graduate studies, he completed his bachelors in electrical engineering at the University of Southern California.

Hamid Javadi (Rice University)
Richard Baraniuk (Rice University)

More from the Same Authors