Unified q-learning framework for mean-field game and control problems

Researchers have developed a unified continuous-time q-learning framework for mean-field game and control problems. This approach, termed the decoupled Iq-function, establishes a martingale characterization that serves as a universal policy evaluation rule for both mean-field game (MFG) and mean-field control (MFC) scenarios. The proposed algorithm is effective even when the environment simulator lacks direct access to population distribution, updating population distribution based on the representative agent's state values. The framework's utility is demonstrated through applications within and beyond the LQ framework, showcasing its efficiency for both MFG and MFC learning tasks. AI

IMPACT Introduces a novel unified q-learning approach for complex game and control problems, potentially advancing reinforcement learning applications.

RANK_REASON The cluster contains a research paper published on arXiv detailing a new theoretical framework and algorithm. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• LQ framework
• Xiaoli Wei

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →