Sequence Training of DNN Acoustic Models With Natural Gradient
This work addresses computational efficiency in speech recognition for researchers and practitioners, though it is incremental as it builds on existing sequence training methods.
The paper tackles the problem of slow convergence in sequence training of DNN acoustic models by proposing a Natural Gradient (NG) optimization framework, which corrects gradients using local KL-divergence curvature and achieves faster convergence than Hessian Free training while maintaining accuracy on a Multi-Genre Broadcast transcription task.
Deep Neural Network (DNN) acoustic models often use discriminative sequence training that optimises an objective function that better approximates the word error rate (WER) than frame-based training. Sequence training is normally implemented using Stochastic Gradient Descent (SGD) or Hessian Free (HF) training. This paper proposes an alternative batch style optimisation framework that employs a Natural Gradient (NG) approach to traverse through the parameter space. By correcting the gradient according to the local curvature of the KL-divergence, the NG optimisation process converges more quickly than HF. Furthermore, the proposed NG approach can be applied to any sequence discriminative training criterion. The efficacy of the NG method is shown using experiments on a Multi-Genre Broadcast (MGB) transcription task that demonstrates both the computational efficiency and the accuracy of the resulting DNN models.