University of Cambridge > Talks.cam > Statistics > On momentum methods and acceleration in stochastic optimization

On momentum methods and acceleration in stochastic optimization

Add to your list(s) Download to your calendar using vCal

  • UserPraneeth Netrapalli (Microsoft Research India)
  • ClockMonday 21 May 2018, 16:00-17:00
  • HouseMR14.

If you have a question about this talk, please contact Quentin Berthet.

It is well known that momentum gradient methods (e.g., Polyak’s heavy ball, Nesterov’s acceleration) yield significant improvements over vanilla gradient descent in deterministic optimization (i.e., where we have access to exact gradient of the function to be minimized). However, there is widespread sentiment that these momentum methods are not effective for the purposes of stochastic optimization due to their instability and error accumulation. Numerous works have attempted to quantify these instabilities in the face of either statistical or non-statistical errors (Paige, 1971; Proakis, 1974; Polyak, 1987; Greenbaum, 1989; Roy and Shynk, 1990; Sharma et al., 1998; d’Aspremont, 2008; Devolder et al., 2013, 2014; Yuan et al., 2016) but a precise understanding is lacking. This work considers these issues for the special case of stochastic approximation for the linear least squares regression problem, and shows that:

1. classical momentum methods (heavy ball and Nesterov’s acceleration) indeed do not offer any improvement over stochastic gradient descent, and

2. introduces an accelerated stochatic gradient method that provably achieves the minimax optimal statistical risk faster than stochastic gradient descent (and classical momentum methods).

Critical to the analysis is a sharp characterization of accelerated stochastic gradient descent as a stochastic process. While the results are rigorously established for the special case of linear least squares regression, experiments suggest that the conclusions hold for the training of deep neural networks.

Joint work with Prateek Jain, Sham M. Kakade, Rahul Kidambi and Aaron Sidford

This talk is part of the Statistics series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2024 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity