Web: http://arxiv.org/abs/2201.10245

Jan. 26, 2022, 2:11 a.m. | Xiaoyu Wang, Mikael Johansson

cs.LG updates on arXiv.org arxiv.org

A theoretical, and potentially also practical, problem with stochastic
gradient descent is that trajectories may escape to infinity. In this note, we
investigate uniform boundedness properties of iterates and function values
along the trajectories of the stochastic gradient descent algorithm and its
important momentum variant. Under smoothness and $R$-dissipativity of the loss
function, we show that broad families of step-sizes, including the widely used
step-decay and cosine with (or without) restart step-sizes, result in uniformly
bounded iterates and function values. …

arxiv uniform variants

More from arxiv.org / cs.LG updates on arXiv.org

Data Analyst, Credit Risk

@ Stripe | US Remote

Senior Data Engineer

@ Snyk | Cluj, Romania, or Remote

Senior Software Engineer (C++), Autonomy Visualization

@ Nuro, Inc. | Mountain View, California (HQ)

Machine Learning Intern (January 2023)

@ Cohere | Toronto, Palo Alto, San Francisco, London

Senior Machine Learning Engineer, Reinforcement Learning, Personalization

@ Spotify | New York, NY

AWS Data Engineer

@ ProCogia | Seattle