Nonconvex Variance Reduced Optimization with Arbitrary Sampling
arXiv:1809.04146
Abstract
We provide the first importance sampling variants of variance reduced algorithms for empirical risk minimization with non-convex loss functions. In particular, we analyze non-convex versions of SVRG, SAGA and SARAH. Our methods have the capacity to speed up the training process by an order of magnitude compared to the state of the art on real datasets. Moreover, we also improve upon current mini-batch analysis of these methods by proposing importance sampling for minibatches in this setting. Surprisingly, our approach can in some regimes lead to superlinear speedup with respect to the minibatch size, which is not usually present in stochastic optimization. All the above results follow from a general analysis of the methods which works with arbitrary sampling, i.e., fully general randomized strategy for the selection of subsets of examples to be sampled in each iteration. Finally, we also perform a novel importance sampling analysis of SARAH in the convex setting.
arXiv comment9 pages, 12 figures, 25 pages of supplementary materials
Cited by in corpus (10)
- Optimal Client Sampling for Federated Learning
- Natural Compression for Distributed Deep Learning
- SGD: General Analysis and Improved Rates
- PAGE: A Simple and Optimal Probabilistic Gradient Estimator for Nonconvex Optimization
- A Better Alternative to Error Feedback for Communication-Efficient Distributed Learning
- L-SVRG and L-Katyusha with Arbitrary Sampling
- 99% of Distributed Optimization is a Waste of Time: The Issue and How to Fix it
- Stochastic Extragradient: General Analysis and Improved Rates
- Effective Proximal Methods for Non-convex Non-smooth Regularized Learning
- Optimization for Supervised Machine Learning: Randomized Algorithms for Data and Parameters