New RL algorithm learns from preferences with unknown link function

Researchers have developed a new reinforcement learning algorithm called Sign-SZPO that can learn from preference feedback even when the relationship between preferences and outcomes is unknown. This approach avoids the restrictive assumption of a known link function, which is common in existing methods. Sign-SZPO estimates the sign of value function differences to construct a policy update direction, demonstrating provable convergence and robustness against mis-specifications in empirical evaluations. AI

IMPACT This research could lead to more robust reinforcement learning systems that can adapt to complex, real-world preference data.

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

Read on arXiv stat.ML →

• alphaXiv
• arXiv
• Bradley--Terry model
• CatalyzeX Code Finder for Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Qining Zhang
• ScienceCast
• Sign-SZPO

• paper
• model release

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