Spectral-Risk Safe Reinforcement Learning with Convergence Guarantees

Neural Information Processing Systems 

[no summary]