Markov abstractions for PAC reinforcement learning in non-Markov decision processes by Ronca, A, Licks, GP, De Giacomo, G