New Bethe free energy formulation enhances active inference capabilities

Researchers have proposed a new formulation for active inference using a Bethe free energy functional, which supports inference by message passing. This approach addresses limitations of existing methods where the free energy functional lacks a Kullback-Leibler structure, hindering message passing treatments. The proposed method imposes an information constraint on the Bethe Lagrangian, ensuring that the mutual information between future observations, states, and parameters given actions is sufficiently large. This constrained Bethe Lagrangian can recover the expected free energy solution under specific conditions, allowing for variations in the agent's epistemic drive. AI

IMPACT This new formulation could improve how agents learn and make decisions by enhancing message passing capabilities in active inference models.

RANK_REASON The cluster contains an academic paper detailing a new theoretical formulation in a subfield of machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• Active Inference
• arXiv
• Bethe free-energy approximations for disordered quantum systems
• Bethe Lagrangian
• Efe
• Karush–Kuhn–Tucker conditions
• Kullback--Leibler divergence
• Q-MDP

• paper
• other

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