Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural... Connectionist Temporal Classi cation: Labelling Unsegmented Sequence Data with Recurrent Neural Networks Alex Graves1 alex@idsia.ch Santiago Fern´ndez1 a santiago@idsia.ch Faustino Gomez1 tino@idsia.ch J rgen Schmidhuber1,2 u juergen@idsia.ch 1 Istituto Dalle Molle di Studi sull ™Intelligenza Arti ciale (IDSIA), Galleria 2, 6928 Manno-Lugano, Switzerland 2 Technische Universit t M nchen (TUM), Boltzmannstr. 3, 85748 Garching, Munich, Germany a u Abstract Many real-world sequence learning tasks require the prediction of sequences of labels from noisy, unsegmented input data. In speech recognition, for example, an acoustic signal is transcribed into words or sub-word units. Recurrent neural networks (RNNs) are powerful sequence learners that would seem well suited to such tasks. However, because they require pre-segmented training data, and post-processing to transform their outputs into label sequences, their applicability has so far been limited. This paper presents a novel method for training RNNs to label unsegmented sequences directly, thereby solving both problems. An experiment on the TIMIT speech corpus demonstrates its advantages over both a baseline HMM and a hybrid HMM-RNN. belling. While these approaches have proved successful for many problems, they have several drawbacks: (1) they usually require a signi cant amount of task speci c knowledge, e.g. to design the http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks

Association for Computing Machinery — Jun 25, 2006

Loading next page...
/lp/association-for-computing-machinery/connectionist-temporal-classification-labelling-unsegmented-sequence-e5cqnJCup0

References (18)

Datasource
Association for Computing Machinery
Copyright
Copyright © 2006 by ACM Inc.
ISBN
1-59593-383-2
doi
10.1145/1143844.1143891
Publisher site
See Article on Publisher Site

Abstract

Connectionist Temporal Classi cation: Labelling Unsegmented Sequence Data with Recurrent Neural Networks Alex Graves1 alex@idsia.ch Santiago Fern´ndez1 a santiago@idsia.ch Faustino Gomez1 tino@idsia.ch J rgen Schmidhuber1,2 u juergen@idsia.ch 1 Istituto Dalle Molle di Studi sull ™Intelligenza Arti ciale (IDSIA), Galleria 2, 6928 Manno-Lugano, Switzerland 2 Technische Universit t M nchen (TUM), Boltzmannstr. 3, 85748 Garching, Munich, Germany a u Abstract Many real-world sequence learning tasks require the prediction of sequences of labels from noisy, unsegmented input data. In speech recognition, for example, an acoustic signal is transcribed into words or sub-word units. Recurrent neural networks (RNNs) are powerful sequence learners that would seem well suited to such tasks. However, because they require pre-segmented training data, and post-processing to transform their outputs into label sequences, their applicability has so far been limited. This paper presents a novel method for training RNNs to label unsegmented sequences directly, thereby solving both problems. An experiment on the TIMIT speech corpus demonstrates its advantages over both a baseline HMM and a hybrid HMM-RNN. belling. While these approaches have proved successful for many problems, they have several drawbacks: (1) they usually require a signi cant amount of task speci c knowledge, e.g. to design the

There are no references for this article.