Incremental Without Replacement Sampling in Nonconvex Optimization
arXiv:2007.07557 · doi:10.1007/s10957-021-01883-2
Abstract
Minibatch decomposition methods for empirical risk minimization are commonly analysed in a stochastic approximation setting, also known as sampling with replacement. On the other hands modern implementations of such techniques are incremental: they rely on sampling without replacement, for which available analysis are much scarcer. We provide convergence guaranties for the latter variant by analysing a versatile incremental gradient scheme. For this scheme, we consider constant, decreasing or adaptive step sizes. In the smooth setting we obtain explicit complexity estimates in terms of epoch counter. In the nonsmooth setting we prove that the sequence is attracted by solutions of optimality conditions of the problem.
arXiv commentJournal of Optimization Theory and Applications, 2021
References in corpus (9)
- Communication-Efficient Learning of Deep Networks from Decentralized Data
- Why Random Reshuffling Beats Stochastic Gradient Descent
- On the Convergence Rate of Incremental Aggregated Gradient Algorithms
- Adaptivity without Compromise: A Momentumized, Adaptive, Dual Averaged Gradient Method for Stochastic Optimization
- Random Reshuffling: Simple Analysis with Vast Improvements
- Stochastic Learning under Random Reshuffling with Constant Step-sizes
- A Unified Convergence Analysis for Shuffling-Type Gradient Methods
- Complexity of Finding Stationary Points of Nonsmooth Nonconvex Functions
- Closing the convergence gap of SGD without replacement