New paper axiomatizes Boltzmann rationality in reinforcement learning

A new paper introduces an axiomatic characterization for Boltzmann rationality, a common model of stochastic choice in reinforcement learning. The research distinguishes between randomness in choice and environmental chance, proposing that by restricting the Independence axiom to environmental lotteries, the Boltzmann policy and its associated soft Bellman equation can be uniquely derived. This framework offers a normative assessment for agent design, clarifying when the Independence of Irrelevant Alternatives axiom is appropriate. AI

IMPACT Provides a theoretical framework for understanding agent decision-making in reinforcement learning.

RANK_REASON The cluster contains a single academic paper published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Boltzmann policy
• Boltzmann Rationality
• economics
• hard Bellman equation
• Independence
• information theory
• Markov decision process
• reinforcement learning
• soft Bellman equation
• softmax policy
• von Neumann–Morgenstern utility theorem

• paper
• other

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