AI Paper Atlas

Research map / Optimization / Nonconvex and smooth optimization

Theoretical Limits of Pipeline Parallel Optimization and Application to Distributed Deep Learning

Igor Colin, Ludovic DOS SANTOS, Kevin Scaman

NeurIPS 2019 · Optimization › Nonconvex and smooth optimization · approach: Coordinate and Adam convergence analysis · 3 citations (OpenAlex)

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
Use randomized smoothing with pipeline-parallel gradient computationPPRS versus GD and AGD · CIFAR10 adversarial attack (frog to horse, AlexNet)loss convergence speedImproved
Increase the number of gradient samples (ablation)PPRS · CIFAR10 adversarial attack (frog to horse, AlexNet)loss convergence speedImproved
Use K=10 instead of K=2 at learning rate 10^{-3} (ablation)PPRS · CIFAR10 adversarial attack (frog to horse, AlexNet)convergence: 10^{-3}Improved
Use GD instead of PPRSCIFAR10 adversarial attack (frog to horse, AlexNet) · CIFAR10 adversarial attack (frog to horse, AlexNet)loss convergence speedHurt
Use AGD instead of PPRSCIFAR10 adversarial attack (frog to horse, AlexNet) · CIFAR10 adversarial attack (frog to horse, AlexNet)loss convergence speedHurt
Use a smaller learning rate for GD or AGD than for PPRSGD and AGD · CIFAR10 adversarial attack (frog to horse, AlexNet)loss convergence speed: 10^{-5}; 10^{-7}Hurt

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