Information-Theoretic Generalization Bounds for SGLD via Data-Dependent Estimates
From MaRDI portal
Publication:6328613
arXiv1911.02151MaRDI QIDQ6328613FDOQ6328613
Authors: Jeffrey Negrea, Mahdi Haghifam, Gintare Karolina Dziugaite, Ashish Khisti, Daniel M. Roy 
Publication date: 5 November 2019
Abstract: In this work, we improve upon the stepwise analysis of noisy iterative learning algorithms initiated by Pensia, Jog, and Loh (2018) and recently extended by Bu, Zou, and Veeravalli (2019). Our main contributions are significantly improved mutual information bounds for Stochastic Gradient Langevin Dynamics via data-dependent estimates. Our approach is based on the variational characterization of mutual information and the use of data-dependent priors that forecast the mini-batch gradient based on a subset of the training samples. Our approach is broadly applicable within the information-theoretic framework of Russo and Zou (2015) and Xu and Raginsky (2017). Our bound can be tied to a measure of flatness of the empirical risk surface. As compared with other bounds that depend on the squared norms of gradients, empirical investigations show that the terms in our bounds are orders of magnitude smaller.
Has companion code repository: https://github.com/jnegrea/neurips2019-5904-code
This page was built for publication: Information-Theoretic Generalization Bounds for SGLD via Data-Dependent Estimates
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6328613)
