Research map / Bandits and online learning / Online convex optimization and games
Trading Off Resource Budgets For Improved Regret Bounds
Thomas Orton, Damon Falck
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 changed | Where | Measured | Effect |
|---|---|---|---|
| increase FPML-partial subroutine budget B-tilde from 1 to 2 in OG_hybrid (B=6) (ablation) | OG_hybrid · Online HPO, B=6 | mean reward: 0.785 -> 0.814 | Improved |
| increase B-tilde from 2 to 3 in OG_hybrid (B=6) (ablation) | OG_hybrid · Online HPO, B=6 | mean reward: 0.814 -> 0.836 | Improved |
| increase B-tilde from 1 to 2 in OG_hybrid (B=4) (ablation) | OG_hybrid · Online HPO, B=4 | mean reward: 0.716 -> 0.756 | Improved |
| replace OG's Exp3 subroutine with FPML-partial(3) in hybrid (B=6) | OG · Online HPO, B=6 | mean reward: 0.767 -> 0.836 | Improved |
| replace OG's Exp3 subroutine with FPML-partial(2) in hybrid (B=4) | OG · Online HPO, B=4 | mean reward: 0.689 -> 0.756 | Improved |
| use FPML-partial(6) instead of OG (B=6) | OG · Online HPO, B=6 | mean reward: 0.767 -> 0.888 | Improved |
| use FPML-partial(4) instead of OG(4) | OG · Online HPO, B=4 | mean reward: 0.689 -> 0.813 | Improved |
| use OG_hybrid(B,1) instead of OG (different single-arm subroutine, FPML-partial(1) vs Exp3) | OG · Online HPO, B=6 | mean reward: 0.767 -> 0.785 | No clear effect |
| use OG_hybrid(4,1) instead of OG(4) | OG · Online HPO, B=4 | mean reward: 0.689 -> 0.716 | No clear effect |
| FPML-partial(6) versus best fixed subset of 6 in hindsight | BIH(6) · Online HPO, B=6 | mean reward: 0.901 -> 0.888 | Hurt |
| FPML-partial(4) versus best fixed subset of 4 in hindsight | BIH(4) · Online HPO, B=4 | mean reward: 0.836 -> 0.813 | Hurt |
| lower budget from B=6 to B=4 for FPML-partial (ablation) | FPML-partial · Online HPO | mean reward: 0.888 -> 0.813 | Hurt |
Closest papers by meaning
- No-Regret Learning Under Adversarial Resource Constraints: A Spending Plan Is All You Need! (NeurIPS 2025 · similarity 0.90)
- Follow-the-Perturbed-Leader Nearly Achieves Best-of-Both-Worlds for the m-Set Semi-Bandit Problems (NeurIPS 2025 · similarity 0.90)
- From Average Sensitivity to Small-Loss Regret Bounds under Random-Order Model (NeurIPS 2026 · similarity 0.88)
- Online Learning under Budget and ROI Constraints via Weak Adaptivity (ICML 2024 · similarity 0.88)
- A Simple and Adaptive Learning Rate for FTRL in Online Learning with Minimax Regret of $\Theta(T^{2/3})$ and its Application to Best-of-Both-Worlds (NeurIPS 2024 · similarity 0.88)
- Regret Bounds for Adversarial Contextual Bandits with General Function Approximation and Delayed Feedback (NeurIPS 2025 · similarity 0.88)
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.
