Distilling Knowledge from Ensembles of Neural Networks for Speech Recognition

  Abstract