Non-Ergodic Convergence Analysis of Heavy-Ball Algorithms

Authors

  • Tao Sun National University of Defense Technology
  • Penghang Yin National University of Defense Technology
  • Dongsheng Li National University of Defense Technology
  • Chun Huang National University of Defense Technology
  • Lei Guan National University of Defense Technology
  • Hao Jiang National University of Defense Technology

DOI:

https://doi.org/10.1609/aaai.v33i01.33015033

Abstract

In this paper, we revisit the convergence of the Heavy-ball method, and present improved convergence complexity results in the convex setting. We provide the first non-ergodic O(1/k) rate result of the Heavy-ball algorithm with constant step size for coercive objective functions. For objective functions satisfying a relaxed strongly convex condition, the linear convergence is established under weaker assumptions on the step size and inertial parameter than made in the existing literature. We extend our results to multi-block version of the algorithm with both the cyclic and stochastic update rules. In addition, our results can also be extended to decentralized optimization, where the ergodic analysis is not applicable.

Downloads

Published

2019-07-17

How to Cite

Sun, T., Yin, P., Li, D., Huang, C., Guan, L., & Jiang, H. (2019). Non-Ergodic Convergence Analysis of Heavy-Ball Algorithms. Proceedings of the AAAI Conference on Artificial Intelligence, 33(01), 5033-5040. https://doi.org/10.1609/aaai.v33i01.33015033

Issue

Section

AAAI Technical Track: Machine Learning