Meta-value Learning: A General Framework For Learning With Learning Awareness

Abstract

Gradient-based learning in multi-agent systems is difficult because the gradient derives from a first-order model which does not account for the interaction between agents' learning processes. LOLA (arXiv:1709.04326) accounts for this by differentiating through one step of optimization. We propose to judge joint policies by their long-term prospects as measured by the meta-value, a discounted sum over the returns of future optimization iterates. We apply a form of Q-learning to the meta-game of optimization, in a way that avoids the need to explicitly represent the continuous action space of policy updates. The resulting method, MeVa, is consistent and far-sighted, and does not require REINFORCE estimators. We analyze the behavior of our method on a toy game and compare to prior work on repeated matrix games.

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keycooijmans2023meta

Related papers

• A Policy Gradient Algorithm For Learning To Learn In Multiagent Reinforcement Learning (2020)0.00
• Learning Meta Representations For Agents In Multi-agent Reinforcement Learning (2021)0.00
• Finding Useful Predictions By Meta-gradient Descent To Improve Decision-making (2021)0.00
• Multi-agent Cooperation Through Learning-aware Policy Gradients (2024)0.00
• Unifying Gradient Estimators For Meta-reinforcement Learning Via Off-policy Evaluation (2021)0.00
• Credit Assignment With Meta-policy Gradient For Multi-agent Reinforcement Learning (2021)0.00
• Meta-gradient Reinforcement Learning With An Objective Discovered Online (2020)0.00
• Qatten: A General Framework For Cooperative Multiagent Reinforcement Learning (2020)0.00