Two-stage ASGD framework for parallel training of DNN acoustic models using Ethernet.

Zhichao Wang, Xingyu Na, Xin Li, Jielin Pan, Yonghong Yan

Browse the full ASRU paper archive.