AI Paper Atlas

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

Online multiclass boosting

Young Hun Jung, Jack Goetz, Ambuj Tewari

NeurIPS 2017 · Bandits and online learning › Online convex optimization and games · approach: Online learning with feedback graphs · 6 citations (OpenAlex)

Read the paper

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 adaptive weighting and expert selection in Adaboost.OLM instead of OnlineMBBM's fixed-edge strategyOnline multiclass boosting · UCI datasets (final 20%)accuracy: Balance: 0.754 vs 0.821; Mice: 0.561 vs 0.695; Cars: 0.930 vs 0.914; Mushroom: 1.000 vs 1.000; Nursery: 0.966 vs 0.969; Mixed
Use Adaboost.OLM instead of the strongest external baselineOnline multiclass boosting · Carsaccuracy: 0.924 -> 0.930Improved
Use Adaboost.OLM instead of the strongest external baselineOnline multiclass boosting · Nurseryaccuracy: 0.953 -> 0.966Improved
Use Adaboost.OLM instead of the strongest external baselineOnline multiclass boosting · ISOLETaccuracy: 0.515 -> 0.521Improved
Use Adaboost.OLM instead of the strongest external baselineOnline multiclass boosting · Movementaccuracy: 0.915 -> 0.962Improved
Use logistic loss rather than exponential loss or square hinge loss (ablation)Adaboost.OLM · Online multiclass boostingmistake boundImproved
Use logistic loss rather than exponential loss (ablation)Adaboost.OLM · Online multiclass boostingweight stabilityImproved

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