Research map / Bandits and online learning / Multi-armed bandits
Regulating Greed Over Time in Multi-Armed Bandits
Stefano Tracà, Cynthia Rudin, Weiyu Yan
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 |
|---|---|---|---|
| regulate greed over time with a hard threshold (epsilon-z greedy) | epsilon-greedy · Yahoo! Webscope (epsilon-greedy, immortal) | mean reward per turn: 0.0406 -> 0.0493 | Improved |
| regulate greed over time softly (Soft epsilon-greedy) | epsilon-greedy · Yahoo! Webscope (epsilon-greedy, immortal) | mean reward per turn: 0.0406 -> 0.0588 | Improved |
| use variable arm pool algorithm instead of standard epsilon-greedy | epsilon-greedy · Yahoo! Webscope (epsilon-greedy, immortal) | mean reward per turn: 0.0406 -> 0.0688 | Improved |
| regulate greed with threshold (UCB-z) | UCB · Yahoo! Webscope (UCB, immortal) | mean reward per turn: 0.0365 -> 0.0441 | Improved |
| regulate greed softly (Soft UCB) | UCB · Yahoo! Webscope (UCB, immortal) | mean reward per turn: 0.0365 -> 0.0396 | Improved |
| use soft greed regulation with mortal arms (UCB soft mortal) instead of standard UCB | UCB · Yahoo! Webscope (UCB, mortal) | mean reward per turn: 0.0523 -> 0.0840 | Improved |
| use UCB soft mortal instead of UCB-L | UCB-L · Yahoo! Webscope (UCB, mortal) | mean reward per turn: 0.0787 -> 0.0840 | Improved |
| regulate greed over time versus smarter UCB and epsilon-greedy (G(t) discounting) | bandit algorithms · simulated Wave, Christmas, Step greed (200 arms, 1500 turns) | final cumulative reward | Improved |
Closest papers by meaning
- Non-Stationary Bandits with Auto-Regressive Temporal Dependency (NeurIPS 2023 · similarity 0.87)
- Unreasonable Effectiveness of Greedy Algorithms in Multi-Armed Bandit with Many Arms (NeurIPS 2020 · similarity 0.87)
- Rebounding Bandits for Modeling Satiation Effects (NeurIPS 2021 · similarity 0.86)
- Causal Bandits: The Pareto Optimal Frontier of Adaptivity, a Reduction to Linear Bandits, and Limitations around Unknown Marginals (ICML 2024 · similarity 0.85)
- Greedy Algorithms for Structured Bandits: A Sharp Characterization of Asymptotic Success / Failure (NeurIPS 2025 · similarity 0.85)
- Mixed-Effects Contextual Bandits (AAAI 2024 · 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.
