AI Paper Atlas

Research map / Bandits and online learning / Online convex optimization and games

Faster Matchings via Learned Duals

Michael Dinitz, Sungjin Im, Thomas Lavastida, Benjamin Moseley, Sergei Vassilvitskii

NeurIPS 2021 · Bandits and online learning › Online convex optimization and games · approach: Learning-augmented online algorithms · 5 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
warm-start Hungarian with learned duals instead of zero dualsHungarian algorithm · type model, batch (l=50, v up to 300)iteration countImproved
warm-start with learned duals under extreme noise (v up to 2^20)Hungarian algorithm · type model, batch, geometric viteration countNo clear effect
warm-start with learned duals on clustering-derived real instancesHungarian algorithm · UCI datasets, k=500, batchiteration count: greater than 2x improvementImproved
learn duals online from previous instancesHungarian algorithm · type model l=100 v=200, onlinerun time / iterationsImproved
learn duals online on KDD and Covertype (k=500)Hungarian algorithm · KDD, Covertype, onlineiterationsImproved
make the instance distribution more concentrated (lower noise variance) (ablation)Learned Duals · type modeliteration countImproved
use a non-concentrated distribution with little to learn (ablation)Learned Duals · type model with high varianceiteration countNo clear effect
add feasibility-recovery step to learned duals (ablation)Learned Duals · runtimeextra running timeNo clear effect

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