notesum.ai
Published at October 22Convex Markov Games: A Framework for Fairness, Imitation, and Creativity in Multi-Agent Learning
cs.GT
cs.AI
cs.MA
Released Date: October 22, 2024
Authors: Ian Gemp1, Andreas Haupt2, Luke Marris1, Siqi Liu1, Georgios Piliouras1
Aff.: 1Google DeepMind, London, UK; 2MIT, Cambridge, MA

| Linear Loss | Convex Loss (Concave Utility) | |
|---|---|---|
| Single-Agent | Markov Decision Process (MDP) | convex MDP |
| Multi-Agent | Markov (Stochastic) Game | (Ours) convex Markov Game |