Multiple parallel hidden layers and other improvements to recurrent neural network language modeling.

Diamantino Caseiro, Andrej Ljolje

Browse the full ICASSP paper archive.