@becker

On the difficulty of training recurrent neural networks

, , and . International Conference on Machine Learning, page 1310--1318. (2013)