Research map / Multi-agent systems and games / Games and equilibria
Decentralized Reinforcement Learning: Global Decision-Making via Local Economic Transactions
Michael Chang, Sid Kaushik, S. Matthew Weinberg, Tom Griffiths, Sergey Levine
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 a cloned society with redundant primitives (ablation) | Vickrey society · Chain | global return | Improved |
| Use the cloned CCV implementation rather than other tested implementations (ablation) | Cloned Vickrey society · Chain | sample efficiency | Improved |
| Add redundant primitives (ablation) | CCV society · Duality | global return | Improved |
| Use cloned CCV society instead of monolithic baselines | Two Rooms · Two Rooms pre-training and transfer tasks | sample efficiency | Improved |
| Use cloned CCV society instead of a hierarchical monolithic baseline | Two Rooms transfer task · Two Rooms | weight change during transfer | Improved |
| Use cloned primitives with independently parameterized bidding policies rather than shared policy weights (ablation) | Cloned society · Market MDP experiments | global performance | No clear effect |
| Use a solitary society rather than a cloned society (ablation) | Vickrey society · Market Bandit | bid accuracy | Hurt |
| Use a solitary society rather than a cloned society (ablation) | Vickrey society · Chain | global return | Hurt |
| Use the Environment Reward objective instead of auction utility (ablation) | Society · Market Bandit | bid accuracy | Hurt |
| Use the first-price auction instead of the cloned Vickrey mechanism (ablation) | Society · Market Bandit | bid accuracy | Hurt |
| Use solitary Vickrey instead of cloned Vickrey (ablation) | Society · Market Bandit | global policy quality | Hurt |
| Use a monolithic learner instead of cloned CCV society | Two Rooms · Two Rooms pre-training and transfer tasks | sample efficiency | Hurt |
Closest papers by meaning
- Decentralized Q-learning in Zero-sum Markov Games (NeurIPS 2021 · similarity 0.91)
- Decentralised Learning in Systems With Many, Many Strategic Agents (AAAI 2018 · similarity 0.90)
- Tractable Multi-Agent Reinforcement Learning through Behavioral Economics (ICLR 2025 · similarity 0.89)
- Economy of Minds: Emerging Multi-Agent Intelligence with Economic Interactions (NeurIPS 2026 · similarity 0.88)
- Impact of Decentralized Learning on Player Utilities in Stackelberg Games (ICML 2024 · similarity 0.88)
- Decomposing Temporal Equilibrium Strategy for Coordinated Distributed Multi-Agent Reinforcement Learning (AAAI 2024 · 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.
