Uncertainty propagation through deep neural networks

Abstract : In order to improve the ASR performance in noisy environments , distorted speech is typically pre-processed by a speech enhancement algorithm, which usually results in a speech estimate containing residual noise and distortion. We may also have some measures of uncertainty or variance of the estimate. Uncertainty decoding is a framework that utilizes this knowledge of uncertainty in the input features during acoustic model scoring. Such frameworks have been well explored for traditional probabilistic models, but their optimal use for deep neural network (DNN)-based ASR systems is not yet clear. In this paper, we study the propagation of observation uncertainties through the layers of a DNN-based acoustic model. Since this is intractable due to the nonlinearities of the DNN, we employ approximate propagation methods, including Monte Carlo sampling , the unscented transform, and the piecewise exponential approximation of the activation function, to estimate the distribution of acoustic scores. Finally, the expected value of the acoustic score distribution is used for decoding, which is shown to further improve the ASR accuracy on the CHiME database, relative to a highly optimized DNN baseline.
Document type :
Conference papers
Complete list of metadatas

https://hal.inria.fr/hal-01162550
Contributor : Emmanuel Vincent <>
Submitted on : Wednesday, June 10, 2015 - 6:06:48 PM
Last modification on : Saturday, March 30, 2019 - 1:26:19 AM
Long-term archiving on : Tuesday, April 25, 2017 - 6:31:56 AM

File

abdelaziz_IS15.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01162550, version 1

Citation

Ahmed H. Abdelaziz, Shinji Watanabe, John R. Hershey, Emmanuel Vincent, Dorothea Kolossa. Uncertainty propagation through deep neural networks. Interspeech 2015, Sep 2015, Dresden, Germany. ⟨hal-01162550⟩

Share

Metrics

Record views

741

Files downloads

1509