Research map / Bandits and online learning / Online convex optimization and games
Online Selective Classification with Limited Feedback
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Randomly abstain at rate p and eliminate classifiers observed making mistakes | VUE · Adaptive adversaries | mistakes: M_T \le \frac{9N \log(2N/\delta)}{p} | Helped |
| Set exploration rate p = \sqrt{N/T} | VUE · Adaptive adversaries | symmetric regret: \max(M_T, A_T - A_T^*) \lesssim \sqrt{NT} \log(N/\delta) | Helped |
| Use the VUE-PROD weighting strategy with exploration rate p | Online selective classification · Stochastic adversaries | mistakes: \mathbb{E}[M_T] \le 8\frac{\log T \log(NT)}{p} | Helped |
| Use weighted PROD loss with uniform exploration | Online selective classification · Adaptive adversaries | mistakes: \mathbb{E}[M_T] \le \frac{2\log N}{ p} + \frac{2\lambda}{p}\mathbb{E}[A_T^*]; symmetric regret bounds scale as T^{2/3} i | Mixed |
| Use the VUE analysis based on context-domain size instead of classifier count | VUE · Adaptive adversaries, finite context domain | mistakes: M_T &\le \frac{9|\mathcal{X}| \log(2|\mathcal{X}|/\delta)}{p} | Helped |
| Allow the PROD prediction distribution to depend on the current context | VUE-PROD · Stochastic adversaries | excess abstentions | No clear effect |
| Replace VUE’s random exploration with a hedge-type weighted-loss scheme | Online selective classification · Adaptive adversaries | symmetric regret: T^{2/3} instead of T^{1/2} | Hurt |
Closest papers by meaning
- Online Classification with Predictions (NeurIPS 2024 · similarity 0.87)
- Bandit-Feedback Online Multiclass Classification: Variants and Tradeoffs (NeurIPS 2024 · similarity 0.87)
- Learning with Selectively Labeled Data from Multiple Decision-makers (ICML 2025 · similarity 0.86)
- What Does It Take to Build a Performant Selective Classifier? (NeurIPS 2025 · similarity 0.86)
- Online Set Learning from Precision and Recall Feedback (NeurIPS 2026 · similarity 0.86)
- Online Linear Classification with Massart Noise (ICML 2025 · similarity 0.85)
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.