The paper “Safety-constrained Reinforcement Learning for MDPs” by Sebastian Junges, Nils Jansen, Christian Dehnert, Ufuk Topcu and Joost-Pieter Katoen has been accepted at TACAS 2016 (acceptance rate 25,7%).  The paper addresses the problem: Given an MDP with a cost structure, synthesize an optimal policy subject to safety constraints. In particular, the costs of actions are not known before they are executed.