Research map / Bandits and online learning / Online convex optimization and games
Structured Dynamic Pricing: Optimal Regret in a Global Shrinkage Model
Rashmi Ranjan Bhuyan, Adel Javanmard, Sungchul Kim, Gourab Mukherjee, Ryan A. Rossi, Tong Yu, Handong Zhao
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 network auto-correlation strength rho from 0.1 to 0.5 (ablation) | PSGD pricing policy · Set-up 2 | regret: rho = 0.1, 0.3, 0.5 | Improved |
| use shrinkage (PSGD) instead of unshrunken pricing policy | PSGD pricing policy · Set-up 2, rho = 0.5 | regret | Improved |
| use shrinkage over network instead of unshrunken policy, long horizon | PSGD pricing policy · Set-up 4 imbalanced, n=1000, T=5000 | relative regret reduction: 50.5%, 50.5%, 48.5% | Improved |
| use shrinkage over network instead of unshrunken policy, short horizon | PSGD pricing policy · Set-up 4 imbalanced, n=1000, T=100 | relative regret reduction: -1.3%, -4.4%, -0.8% | Hurt |
| increase imbalance of sampling design from 0.7 to 0.9 (ablation) | PSGD pricing policy · Set-up 4, n=1000, T=100 vs balanced unshrunken | relative regret reduction: -6.6% -> -16.3% | Hurt |
| use PSGD under imbalanced design against unshrunken policy in balanced design, long horizon | PSGD pricing policy · Set-up 4, n=1000, T=5000 | relative regret reduction: 47.9%, 44.9%, 35.8% | Improved |
| use PSGD in network built from demographic variables only | PSGD pricing policy · Set-up 5, n=1000, T=500 | relative regret reduction: -13.5%, -18.0%, -20.8% | Hurt |
| use PSGD in network built from demographic variables only, long horizon | PSGD pricing policy · Set-up 5, n=1000, T=5000 | relative regret reduction: 54.8%, 54.3%, 55.7% | Improved |
| use PSGD in network built from economic variables only | PSGD pricing policy · Set-up 6, n=1000, T=5000 | relative regret reduction: 29.3%, 27.7%, 29.1% | Improved |
| low-lead segments are well connected in the network instead of poorly connected (ablation) | PSGD pricing policy · Set-up 7, 10 low-lead states | relative regret vs unshrunken | Improved |
Closest papers by meaning
- Transfer Faster, Price Smarter: Minimax Dynamic Pricing under Cross-Market Preference Shift (NeurIPS 2025 · similarity 0.89)
- Improved Algorithms for Contextual Dynamic Pricing (NeurIPS 2024 · similarity 0.88)
- Near-Optimal Regret-Queue Length Tradeoff in Online Learning for Two-Sided Markets (NeurIPS 2025 · similarity 0.87)
- Learning to price with resource constraints: from full information to machine-learned prices (NeurIPS 2025 · similarity 0.87)
- Transfer Learning for Nonparametric Contextual Dynamic Pricing (ICML 2025 · similarity 0.87)
- Optimal Contextual Pricing under Agnostic Non-Lipschitz Demand (NeurIPS 2026 · similarity 0.87)
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.
