A rigorous foundation for provable agentic reasoning. Establishes a trust layer for neurosymbolic AI by formalizing optimization constraints in Lean 4. Replaces empirical approximation with deductive verification, ensuring agentic behaviors adhere to strict safety bounds.
program-synthesis convex-optimization mathlib formal-logic neural-symbolic lean4 certifiable-robustness chain-of-thought neurosymbolic-ai formal-reasoning autoformalization agentic-systems grounded-reasoning theoretical-ml trustworthy-nlp provable-ai trustworthy-ml mechanized-math mechanized-reasoning
-
Updated
Jan 31, 2026 - Lean