ICML 2019oral25 citations

PA-GD: On the Convergence of Perturbed Alternating Gradient Descent to Second-Order Stationary Points for Structured Nonconvex Optimization

Songtao Lu, Mingyi Hong, Zhengdao Wang

Abstract

Alternating gradient descent (A-GD) is a simple but popular algorithm in machine learning, which updates two blocks of variables in an alternating manner using gradient descent steps. In this paper, we consider a smooth unconstrained nonconvex optimization problem, and propose a perturbed A-GD (PA-GD) which is able to converge (with high probability) to the second-order stationary points (SOSPs) with a global sublinear rate. Existing analysis on A-GD type algorithm either only guarantees convergence to first-order solutions, or converges to second-order solutions asymptotically (without rates). To the best of our knowledge, this is the first alternating type algorithm that takes $\mathcal{O}(\text{polylog}(d)/\epsilon^2)$ iterations to achieve an ($\epsilon,\sqrt{\epsilon}$)-SOSP with high probability, where polylog$(d)$ denotes the polynomial of the logarithm with respect to problem dimension $d$.

BibTeX
@InProceedings{pmlr-v97-lu19a,
  title = 	 {{PA}-{GD}: On the Convergence of Perturbed Alternating Gradient Descent to Second-Order Stationary Points for Structured Nonconvex Optimization},
  author =       {Lu, Songtao and Hong, Mingyi and Wang, Zhengdao},
  booktitle = 	 {Proceedings of the 36th International Conference on Machine Learning},
  pages = 	 {4134--4143},
  year = 	 {2019},
  editor = 	 {Chaudhuri, Kamalika and Salakhutdinov, Ruslan},
  volume = 	 {97},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {09--15 Jun},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v97/lu19a/lu19a.pdf},
  url = 	 {https://proceedings.mlr.press/v97/lu19a.html},
  abstract = 	 {Alternating gradient descent (A-GD) is a simple but popular algorithm in machine learning, which updates two blocks of variables in an alternating manner using gradient descent steps. In this paper, we consider a smooth unconstrained nonconvex optimization problem, and propose a perturbed A-GD (PA-GD) which is able to converge (with high probability) to the second-order stationary points (SOSPs) with a global sublinear rate. Existing analysis on A-GD type algorithm either only guarantees convergence to first-order solutions, or converges to second-order solutions asymptotically (without rates). To the best of our knowledge, this is the first alternating type algorithm that takes $\mathcal{O}(\text{polylog}(d)/\epsilon^2)$ iterations to achieve an ($\epsilon,\sqrt{\epsilon}$)-SOSP with high probability, where polylog$(d)$ denotes the polynomial of the logarithm with respect to problem dimension $d$.}
}
PA-GD: On the Convergence of Perturbed Alternating Gradient Descent to Second-Order Stationary Points for Structured Nonconvex Optimization · ICML 2019