A necessary and sufficient stability notion for adaptive generalization
Machine Learning
2020-01-01 v2 Machine Learning
Abstract
We introduce a new notion of the stability of computations, which holds under post-processing and adaptive composition. We show that the notion is both necessary and sufficient to ensure generalization in the face of adaptivity, for any computations that respond to bounded-sensitivity linear queries while providing accuracy with respect to the data sample set. The stability notion is based on quantifying the effect of observing a computation's outputs on the posterior over the data sample elements. We show a separation between this stability notion and previously studied notion and observe that all differentially private algorithms also satisfy this notion.
Cite
@article{arxiv.1906.00930,
title = {A necessary and sufficient stability notion for adaptive generalization},
author = {Katrina Ligett and Moshe Shenfeld},
journal= {arXiv preprint arXiv:1906.00930},
year = {2020}
}