Safe reinforcement learning via formal methods
Safety verification for deep neural networks with provable guarantees
