atlas

Research map / Optimization / Nonconvex and smooth optimization

Proximal Stochastic Methods for Nonsmooth Nonconvex Finite-Sum Optimization

NeurIPS 2016 · Optimization › Nonconvex and smooth optimization · approach: Proximal finite-sum methods · 107 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 proximal SGD with variance-reduced proximal SVRG/SAGA using constant minibatchesnonconvex nonsmooth finite-sum optimization · NN-PCA (rcv1, a9a, mnist, aloi)objective suboptimality vs effective passesHelped
use minibatch b=n^{2/3} in PROXSVRG/PROXSAGA rather than b=1PROXSVRG and PROXSAGA · theory, IFO complexityIFO complexity to eps-accuracy: O(n/eps) to O(n + n^{2/3}/eps)Helped
use constant minibatch b=1 in PROXSVRG (Theorem 1) (ablation)PROXSVRG · theory, IFO and PO complexityIFO and PO complexity: O(n/eps) same as PROXGD IFO; worse PONo clear effect
add PL-condition restarts (PL-SVRG/PL-SAGA) for mu-PL functionsPROXSVRG and PROXSAGA · theory, PL functionsIFO complexity for linear convergence: O(n log(1/eps)) of PROXGD to O((n+n^{2/3}) log(1/eps))Helped

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