M. Vinzent, H. Hermanns, and J. Hoffmann, “Probabilistic Safety Verification of Neural Policies via Predicate Abstraction”, AAAI, vol. 40, no. 43, pp. 36438–36447, Mar. 2026.