A learning based approach to control synthesis of Markov decision processes for linear temporal logic specifications | IEEE Conference Publication | IEEE Xplore