High Probability Guarantees for Nonconvex Stochastic Gradient Descent with Heavy Tails

Shaojie Li, Yong Liu
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:12931-12963, 2022.

Abstract

Stochastic gradient descent (SGD) is the workhorse in modern machine learning and data-driven optimization. Despite its popularity, existing theoretical guarantees for SGD are mainly derived in expectation and for convex learning problems. High probability guarantees of nonconvex SGD are scarce, and typically rely on “light-tail” noise assumptions and study the optimization and generalization performance separately. In this paper, we develop high probability bounds for nonconvex SGD with a joint perspective of optimization and generalization performance. Instead of the light tail assumption, we consider the gradient noise following a heavy-tailed sub-Weibull distribution, a novel class generalizing the sub-Gaussian and sub-Exponential families to potentially heavier-tailed distributions. Under these complicated settings, we first present high probability bounds with best-known rates in general nonconvex learning, then move to nonconvex learning with a gradient dominance curvature condition, for which we improve the learning guarantees to fast rates. We further obtain sharper learning guarantees by considering a mild Bernstein-type noise condition. Our analysis also reveals the effect of trade-offs between the optimization and generalization performance under different conditions. In the last, we show that gradient clipping can be employed to remove the bounded gradient-type assumptions. Additionally, in this case, the stepsize of SGD is completely oblivious to the knowledge of smoothness.

Cite this Paper


BibTeX
@InProceedings{pmlr-v162-li22q, title = {High Probability Guarantees for Nonconvex Stochastic Gradient Descent with Heavy Tails}, author = {Li, Shaojie and Liu, Yong}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {12931--12963}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/li22q/li22q.pdf}, url = {https://proceedings.mlr.press/v162/li22q.html}, abstract = {Stochastic gradient descent (SGD) is the workhorse in modern machine learning and data-driven optimization. Despite its popularity, existing theoretical guarantees for SGD are mainly derived in expectation and for convex learning problems. High probability guarantees of nonconvex SGD are scarce, and typically rely on “light-tail” noise assumptions and study the optimization and generalization performance separately. In this paper, we develop high probability bounds for nonconvex SGD with a joint perspective of optimization and generalization performance. Instead of the light tail assumption, we consider the gradient noise following a heavy-tailed sub-Weibull distribution, a novel class generalizing the sub-Gaussian and sub-Exponential families to potentially heavier-tailed distributions. Under these complicated settings, we first present high probability bounds with best-known rates in general nonconvex learning, then move to nonconvex learning with a gradient dominance curvature condition, for which we improve the learning guarantees to fast rates. We further obtain sharper learning guarantees by considering a mild Bernstein-type noise condition. Our analysis also reveals the effect of trade-offs between the optimization and generalization performance under different conditions. In the last, we show that gradient clipping can be employed to remove the bounded gradient-type assumptions. Additionally, in this case, the stepsize of SGD is completely oblivious to the knowledge of smoothness.} }
Endnote
%0 Conference Paper %T High Probability Guarantees for Nonconvex Stochastic Gradient Descent with Heavy Tails %A Shaojie Li %A Yong Liu %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-li22q %I PMLR %P 12931--12963 %U https://proceedings.mlr.press/v162/li22q.html %V 162 %X Stochastic gradient descent (SGD) is the workhorse in modern machine learning and data-driven optimization. Despite its popularity, existing theoretical guarantees for SGD are mainly derived in expectation and for convex learning problems. High probability guarantees of nonconvex SGD are scarce, and typically rely on “light-tail” noise assumptions and study the optimization and generalization performance separately. In this paper, we develop high probability bounds for nonconvex SGD with a joint perspective of optimization and generalization performance. Instead of the light tail assumption, we consider the gradient noise following a heavy-tailed sub-Weibull distribution, a novel class generalizing the sub-Gaussian and sub-Exponential families to potentially heavier-tailed distributions. Under these complicated settings, we first present high probability bounds with best-known rates in general nonconvex learning, then move to nonconvex learning with a gradient dominance curvature condition, for which we improve the learning guarantees to fast rates. We further obtain sharper learning guarantees by considering a mild Bernstein-type noise condition. Our analysis also reveals the effect of trade-offs between the optimization and generalization performance under different conditions. In the last, we show that gradient clipping can be employed to remove the bounded gradient-type assumptions. Additionally, in this case, the stepsize of SGD is completely oblivious to the knowledge of smoothness.
APA
Li, S. & Liu, Y.. (2022). High Probability Guarantees for Nonconvex Stochastic Gradient Descent with Heavy Tails. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:12931-12963 Available from https://proceedings.mlr.press/v162/li22q.html.

Related Material