This paper introduces a new framework for risk-averse multi-population mean-field games, addressing the uncertainty in other populations' behaviors. The proposed approach optimizes a worst-case expected reward over ambiguity sets of mean-field flows. The authors establish theoretical properties, including the existence of a novel risk-averse equilibrium, and derive contractivity results for learning this equilibrium. A risk-averse fictitious-play scheme is also presented, demonstrating decay to zero exploitability. AI
RANK_REASON The cluster contains a research paper published on arXiv. [lever_c_demoted from research: ic=1 ai=0.4]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →