A context-sensitive-chunk BPTT approach to training deep LSTM/BLSTM recurrent neural networks for offline handwriting recognition.

Kai Chen, Zhi-Jie Yan, Qiang Huo

Browse the full ICDAR paper archive.