Research map / Reinforcement learning / MDPs and planning
NeurWIN: Neural Whittle Index Network For Restless Bandits Via Deep RL
Khaled Nakhleh, Santosh Ganji, Ping-Chun Hsieh, I-Hong Hou, Srinivas Shakkottai
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 |
|---|---|---|---|
| use smaller network (8,14 hidden) instead of 16,32 (ablation) | NeurWIN · Deadline scheduling (4,1) | episodes to converge: 600 -> 1380 | Hurt |
| use larger network (48,64 hidden) instead of 16,32 (ablation) | NeurWIN · Deadline scheduling (4,1) | episodes to converge: 600 -> 180 | Improved |
| use smaller network (8,14 hidden) in wireless scheduling (ablation) | NeurWIN · Wireless scheduling (4,1) | episodes to converge: 5,000 -> 10,000 | Hurt |
| use smaller network in wireless scheduling (10,1) (ablation) | NeurWIN · Wireless scheduling (10,1) | reward vs baseline | Hurt |
| use smaller network in recovering bandits (ablation) | NeurWIN · Recovering bandits (4,1),(10,1) | reward vs baseline | Hurt |
| use noisy simulator with RMS error up to 40% of rewards | NeurWIN · three restless bandit problems N=100, M=25 | discounted reward | No clear effect |
| learn Whittle index with NeurWIN instead of QWIC/WIBQL | QWIC and WIBQL · Deadline scheduling | reward | Improved |
| learn Whittle index with NeurWIN instead of end-to-end deep RL | REINFORCE, WOLP-DDPG, AQL · Deadline scheduling | reward: 2,000 -> 600 | Improved |
Closest papers by meaning
- Reinforcement learning with combinatorial actions for coupled restless bandits (ICLR 2025 · similarity 0.88)
- Provably Efficient Reinforcement Learning for Adversarial Restless Multi-Armed Bandits with Unknown Transitions and Bandit Feedback (ICML 2024 · similarity 0.88)
- Global Rewards in Restless Multi-Armed Bandits (NeurIPS 2024 · similarity 0.85)
- Stochastic Bandits with ReLU Neural Networks (ICML 2024 · similarity 0.85)
- The Bandit Whisperer: Communication Learning for Restless Bandits (AAAI 2025 · similarity 0.85)
- No-Regret Bandit Exploration based on Soft Tree Ensemble Model (NeurIPS 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.
