#artificial intelligence
Aug 2026
Diagnosing Faults in Reinforcement Learning Simulators and World Models with Canonical Polynomial Invariants
This work develops two procedures: screening, which identifies the violated physical constraint, and attribution, which recovers the faulty invariant and identifies the responsible physical parameter, and introduces normal-form deflation and quotient-space recovery.
Zemuy Tesfay Gebrekidan, Hadush Hailu Gebrerufael
· arXiv.org · 0 citations