TY  - RPRT
TI  - Learning over Forward-Invariant Policy Classes: Reinforcement Learning without Safety Concerns
AU  - Chieh Tsai
AU  - Muhammad Junayed Hasan Zahed
AU  - Salim Hariri
AU  - Hossein Rastgoftar
PY  - 2026
UR  - https://arxiv.org/abs/2604.07875
ID  - 2604.07875
ER  -