Non-linear Gradient Mappings and Stochastic Optimization: a General Framework with Applications to Heavy-Tail Noise
Creators
- 1. University of Novi Sad, Faculty of Sciences, Department of Mathematics and Informatics
- 2. niversity of Novi Sad, Faculty of Technical Sciences, Department of Power, Electronic and Communication Engineering
- 3. Amazon Alexa AI
- 4. Department of Electrical and Computer Engineering, Carnegie Mellon University
Description
We introduce a general framework for nonlinear stochastic gradient descent (SGD) for the scenarios when gradient noise exhibits heavy tails. The proposed framework subsumes several popular nonlinearity choices, like clipped, normalized, signed or quantized gradient, but we also consider novel nonlinearity choices. We establish for the considered class of methods strong convergence guarantees assuming a strongly convex cost function with Lipschitz continuous gradients under very general assumptions on the gradient noise. Most notably, we show that, for a nonlinearity with bounded outputs and for the gradient noise that may not have finite moments of order greater than one, the nonlinear SGD’s mean squared error (MSE), or equivalently, the expected cost function’s optimality gap, converges to zero at rate O(1/tζ ), ζ ∈ (0, 1). In contrast, for the same noise setting, the linear SGD generates a sequence with unbounded variances. Furthermore, for general nonlinearities that can be decoupled component wise and a class of joint nonlinearities, we show that the nonlinear SGD asymptotically (locally) achieves a O(1/t) rate in the weak convergence sense and explicitly quantify the corresponding asymptotic variance. Experiments show that, while our framework is more general than existing studies of SGD under heavy-tail noise, several easy-to-implement nonlinearities from our framework are competitive with state-of-the-art alternatives on real data sets with heavy tail noises.
Files
Jakovetic_etal_SIAM2022.pdf
Files
(2.1 MB)
Name | Size | Download all |
---|---|---|
md5:d0526628ef5e32b9ba23b626b2d37a16
|
2.1 MB | Preview Download |