A partially recurrent gating network approach to learning action selection by reinforcement | IEEE Conference Publication | IEEE Xplore