ICML 2021 accepted papers on Multi-agent systems and games
42 of the 1,184 papers accepted at ICML 2021 (3.5%) are on Multi-agent systems and games, compared with 2.9% at ICML 2020. They fall into 3 topics.
See ICML 2021 in the interactive map
Submitting to ICML? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Games and equilibria · 16 papers
- SCC: an efficient deep reinforcement learning agent mastering the game of StarCraft II (30 citations)
- A Sharp Analysis of Model-based Reinforcement Learning with Self-Play (25 citations)
- DouZero: Mastering DouDizhu with Self-Play Deep Reinforcement Learning (21 citations)
- Learning in Nonzero-Sum Stochastic Games with Potentials (10 citations)
- Adversarial Policy Learning in Two-player Competitive Games (7 citations)
- From Local to Global Norm Emergence: Dissolving Self-reinforcing Substructures with Incremental Social Instruments (6 citations)
- Global Convergence of Policy Gradient for Linear-Quadratic Mean-Field Control/Game in Continuous Time (6 citations)
- A New Formalism, Method and Open Issues for Zero-Shot Coordination (4 citations)
- Multi-Agent Training beyond Zero-Sum with Correlated Equilibrium Meta-Solvers (3 citations)
- One for One, or All for All: Equilibria and Optimality of Collaboration in Federated Learning (3 citations)
- Alternative Microfoundations for Strategic Classification (2 citations)
- Incentivizing Compliance with Algorithmic Instruments (2 citations)
and 4 more in the interactive map.
Cooperative multi-agent RL · 14 papers
- Cooperative Exploration for Multi-Agent Deep Reinforcement Learning (31 citations)
- Scalable Evaluation of Multi-Agent Reinforcement Learning with Melting Pot (23 citations)
- Scaling Multi-Agent Reinforcement Learning with Selective Parameter Sharing (17 citations)
- Learning Fair Policies in Decentralized Cooperative Multi-Agent Reinforcement Learning (15 citations)
- Randomized Entity-wise Factorization for Multi-Agent Reinforcement Learning (15 citations)
- Coach-Player Multi-agent Reinforcement Learning for Dynamic Team Composition (12 citations)
- DFAC Framework: Factorizing the Value Function via Quantile Mixture for Multi-Agent Distributional Q-Learning (8 citations)
- UneVEn: Universal Value Exploration for Multi-Agent Reinforcement Learning (7 citations)
- Continuous Coordination As a Realistic Scenario for Lifelong Learning (6 citations)
- Parallel Droplet Control in MEDA Biochips using Multi-Agent Reinforcement Learning (6 citations)
- Tesseract: Tensorised Actors for Multi-Agent Reinforcement Learning (5 citations)
- Towards Open Ad Hoc Teamwork Using Graph-based Policy Learning (4 citations)
and 2 more in the interactive map.
Human-AI and goal-conditioned · 12 papers
- Trajectory Diversity for Zero-Shot Coordination (18 citations)
- Reinforcement Learning Under Moral Uncertainty (11 citations)
- Causal Curiosity: RL Agents Discovering Self-supervised Experiments for Causal Representation Learning (10 citations)
- Navigation Turing Test (NTT): Learning to Evaluate Human-Like Navigation (7 citations)
- Few-shot Language Coordination by Modeling Theory of Mind (6 citations)
- Megaverse: Simulating Embodied Agents at One Million Experiences per Second (4 citations)
- Towards Distraction-Robust Active Visual Tracking (4 citations)
- Learning Interaction Kernels for Agent Systems on Riemannian Manifolds (2 citations)
- Off-Belief Learning (2 citations)
- AGENT: A Benchmark for Core Psychological Reasoning (1 citation)
- Targeted Data Acquisition for Evolving Negotiation Agents
- Value Alignment Verification
