Research map / Bandits and online learning / Online convex optimization and games
Optimizing Urban Service Allocation with Time-Constrained Restless Bandits
Yi Mao, Andrew Perrault
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 |
|---|---|---|---|
| add Whittle-index lookahead scheduling instead of a reward-oblivious IP scheduler within fixed windows (ablation) | RMAB inspection scheduler · Synthetic and CDPH, exactly once per year | total reward | Improved |
| add window optimization on top of schedule optimization (ablation) | RMAB inspection scheduler · Synthetic and CDPH, exactly once per year | total reward | Improved |
| optimize windows but keep the reward-oblivious IP scheduler (ablation) | RMAB inspection scheduler · Synthetic and CDPH, exactly once per year | total reward | Hurt |
| allow some establishments to be inspected twice per year (budget 10-15%) (ablation) | (Opt, Opt, [1,2]) · Synthetic and CDPH | total reward: up to 24% (synthetic), 33% (CDPH) | Improved |
| raise inspection budget from 12% to 15% (ablation) | (Opt, Opt, [1,2]) on CDPH · CDPH | total reward | No clear effect |
| add Gaussian noise to transition parameters, using Opt-Opt instead of Opt-IP | (Opt, Opt, =1) · Synthetic robustness | reward change: -83.69 vs -101.14 at N(0, 0.05) | Improved |
| add 1% surprise inspections with window and schedule optimization | (Opt, Opt, =1) · Synthetic | reward loss: 2.6% -> 1.9% | Improved |
| train neural net on transition probabilities instead of predicting failures with XGBoost | inspection failure prediction · CDPH data | AUC: 0.67 -> 0.75 | Improved |
Closest papers by meaning
- Active Geospatial Search for Efficient Tenant Eviction Outreach (AAAI 2025 · similarity 0.81)
- Optimizing Vital Sign Monitoring in Resource-Constrained Maternal Care: An RL-Based Restless Bandit Approach (AAAI 2025 · similarity 0.80)
- A Decision-Language Model (DLM) for Dynamic Restless Multi-Armed Bandit Tasks in Public Health (NeurIPS 2024 · similarity 0.80)
- Algorithm for Contextual Queueing Bandits with Rate-Optimal Queue Length Regret (NeurIPS 2026 · similarity 0.80)
- Context in Public Health for Underserved Communities: A Bayesian Approach to Online Restless Bandits (AAAI 2025 · similarity 0.79)
- PROBE: Learning to Audit Policy Compliance in Tool-Using LLM Agents (NeurIPS 2026 · similarity 0.79)
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.
