AI Paper Atlas

Research map / Optimization / Optimizers

Without-Replacement Sampling for Stochastic Gradient Methods

Ohad Shamir

NeurIPS 2016 · Optimization › Optimizers · approach: Stochastic gradient MCMC

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Replace with-replacement sampling by without-replacement (single random permutation pass) in SGD for convex Lipschitz losses with O(sqrt(T)) regret algorithmsSGD / online convex learners on linear predictors · Theoretical bound (convex Lipschitz linear predictors)expected suboptimality rateNo clear effect
Use without-replacement sampling in SGD with strongly convex and smooth objectives (step size 2/(lambda t))SGD · Theoretical bound (lambda-strongly convex, smooth losses)expected suboptimality rateNo clear effect
Use without-replacement SVRG (single shuffle, no reshuffling per epoch) for regularized least squaresSVRG · Regularized least squares (theoretical)iterations/passes to reach epsilon accuracyNo clear effect
Simulate without-replacement SVRG on randomly partitioned data across machines instead of communicating individual data points (Lee et al. approach)Distributed SVRG · Distributed regularized least squaresregime of lambda where communication-efficient (lambda ~ 1/m vs 1/sqrt(m))Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy