An Empirical Comparison of Neural Architectures for Reinforcement Learning in Partially Observable Environments

This paper explores the performance of fitted neural Q iteration for reinforcement learning in several partially observable environments, using three recurrent neural network architectures: Long ShortTerm Memory [7], Gated Recurrent Unit [3] and MUT1, a recurrent neural architecture evolved from a pool of several thousands candidate architectures [8]. A variant of fitted Q iteration, based on Advantage values [6, 1] instead of Q values, is also explored. The results show that GRU performs significantly better than LSTM and MUT1 for most of the problems considered, requiring less training episodes and less CPU time before learning a very good policy. Advantage learning also tends to produce better results.