Trainify: A CEGAR-Driven Training and Verification Framework for Safe Deep Reinforcement Learning
Authors: Peng Jin, Jiaxu Tian, Dapeng Zhi, Xuejun Wen and Min Zhang
Paper Information
| Title: | Trainify: A CEGAR-Driven Training and Verification Framework for Safe Deep Reinforcement Learning |
| Authors: | Peng Jin, Jiaxu Tian, Dapeng Zhi, Xuejun Wen and Min Zhang |
| Proceedings: | CAV 2022 All Papers |
| Editors: | Yakir Vizel, Sharon Shoham and Hari Govind Vediramana Krishnan |
| Keywords: | Deep reinforcement learning, model checking, abstraction and refinement, correct-by-construction |
| Pages: | 21 |
| Talk: | Aug 07 15:00 (Session 90A: Formal Methods for Neural Networks) |
| Paper: |  |