Michael Kearns

The Ethical Algorithm: The Science of Socially Aware Algorithm Design

Perspectives on Science and Christian Faith · 2021 · 266 citations

THE ETHICAL ALGORITHM: The Science of Socially Aware Algorithm Design by Michael Kearns and Aaron Roth. New York: Oxford University Press, 2019. 232 pages. Hardcover; $24.95. ISBN: 9780190948207. *Can an algorithm be ethical? That question appears to be similar to asking if a hammer can be ethical. Isn't the ethics solely related to how the hammer is used? Using it to build a house seems ethical; using it to harm another person would be immoral. *That line of thinking would be appropriate if the…

AI model disgorgement: Methods and choices

Proceedings of the National Academy of Sciences · 2024-04-19 · 9 citations

Over the past few years, machine learning models have significantly increased in size and complexity, especially in the area of generative AI such as large language models. These models require massive amounts of data and compute capacity to train, to the extent that concerns over the training data (such as protected or private content) cannot be practically addressed by retraining the model "from scratch" with the questionable data removed or altered. Furthermore, despite significant efforts an…

Scalable Membership Inference Attacks via Quantile Regression

arXiv (Cornell University) · 2023-07-07 · 8 citations

Membership inference attacks are designed to determine, using black box access to trained models, whether a particular example was used in training or not. Membership inference can be formalized as a hypothesis testing problem. The most effective existing attacks estimate the distribution of some test statistic (usually the model's confidence on the true label) on points that were (and were not) used in training by training many \emph{shadow models} -- i.e. models of the same architecture as the…

Reconstruction Attacks on Machine Unlearning: Simple Models are Vulnerable

2024-01-01 · 5 citations

Reconstruction Attacks on Machine Unlearning: Simple Models are Vulnerable

arXiv (Cornell University) · 2024-05-30 · 4 citations

Machine unlearning is motivated by desire for data autonomy: a person can request to have their data's influence removed from deployed models, and those models should be updated as if they were retrained without the person's data. We show that, counter-intuitively, these updates expose individuals to high-accuracy reconstruction attacks which allow the attacker to recover their data in its entirety, even when the original models are so simple that privacy risk might not otherwise have been a con…