Dong Yu, Li Deng, and George E. Dahl
Recently, deep learning techniques have been successfully applied to automatic speech recognition tasks -- first to phonetic recognition with context-independent deep belief network (DBN) hidden Markov models (HMMs) and later to large vocabulary continuous speech recognition using context-dependent (CD) DBN-HMMs. In this paper, we report our most recent experiments designed to understand the roles of the two main phases of the DBN learning -- pre-training and fine tuning -- in the recognition performance of a CD-DBN-HMM based large-vocabulary speech recognizer. As expected, we show that pre-training can initialize weights to a point in the space where fine-tuning can be effective and thus is crucial in training deep structured models. However, a moderate increase of the amount of unlabeled pre-training data has an insignificant effect on the final recognition results as long as the original training size is sufficiently large to initialize the DBN weights. On the other hand, with additional labeled training data, the fine-tuning phase of DBN training can significantly improve the recognition accuracy.
In NIPS 2010 workshop on Deep Learning and Unsupervised Feature Learning