Learning natural language understanding systems from unaligned labels for voice command in smart homes
- Author
- Anastasiia Mishakova, François Portet, Thierry Desot (UGent) and Michel Vacher
- Organization
- Abstract
- Voice command smart home systems have become a target for the industry to provide more natural human computer interaction. To interpret voice command, systems must be able to extract the meaning from natural language; this task is called Natural Language Understanding (NLU). Modern NLU is based on statistical models which are trained on data. However, a current limitation of most NLU statistical models is the dependence on large amount of textual data aligned with target semantic labels. This is highly time-consuming. Moreover, they require training several separate models for predicting intents, slot-labels and slot-values. In this paper, we propose to use a sequence-to-sequence neural architecture to train NLU models which do not need aligned data and can jointly learn the intent, slot-label and slot-value prediction tasks. This approach has been evaluated both on a voice command dataset we acquired for the purpose of the study as well as on a publicly available dataset. The experiments show that a single model learned on unaligned data is competitive with state-of-the-art models which depend on aligned data.
- Keywords
- Natural Language Understanding, Smart Environments, Deep Neural Network, Voice-User Interface, AUDIO
Downloads
-
(...).pdf
- full text (Published version)
- |
- UGent only
- |
- |
- 306.32 KB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-8750176
- MLA
- Mishakova, Anastasiia, et al. “Learning Natural Language Understanding Systems from Unaligned Labels for Voice Command in Smart Homes.” 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS), IEEE, 2019, pp. 832–37, doi:10.1109/PERCOMW.2019.8730721.
- APA
- Mishakova, A., Portet, F., Desot, T., & Vacher, M. (2019). Learning natural language understanding systems from unaligned labels for voice command in smart homes. 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS), 832–837. https://doi.org/10.1109/PERCOMW.2019.8730721
- Chicago author-date
- Mishakova, Anastasiia, François Portet, Thierry Desot, and Michel Vacher. 2019. “Learning Natural Language Understanding Systems from Unaligned Labels for Voice Command in Smart Homes.” In 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS), 832–37. IEEE. https://doi.org/10.1109/PERCOMW.2019.8730721.
- Chicago author-date (all authors)
- Mishakova, Anastasiia, François Portet, Thierry Desot, and Michel Vacher. 2019. “Learning Natural Language Understanding Systems from Unaligned Labels for Voice Command in Smart Homes.” In 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS), 832–837. IEEE. doi:10.1109/PERCOMW.2019.8730721.
- Vancouver
- 1.Mishakova A, Portet F, Desot T, Vacher M. Learning natural language understanding systems from unaligned labels for voice command in smart homes. In: 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS). IEEE; 2019. p. 832–7.
- IEEE
- [1]A. Mishakova, F. Portet, T. Desot, and M. Vacher, “Learning natural language understanding systems from unaligned labels for voice command in smart homes,” in 2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS), Kyoto, JAPAN, 2019, pp. 832–837.
@inproceedings{8750176,
abstract = {{Voice command smart home systems have become a target for the industry to provide more natural human computer interaction. To interpret voice command, systems must be able to extract the meaning from natural language; this task is called Natural Language Understanding (NLU). Modern NLU is based on statistical models which are trained on data. However, a current limitation of most NLU statistical models is the dependence on large amount of textual data aligned with target semantic labels. This is highly time-consuming. Moreover, they require training several separate models for predicting intents, slot-labels and slot-values. In this paper, we propose to use a sequence-to-sequence neural architecture to train NLU models which do not need aligned data and can jointly learn the intent, slot-label and slot-value prediction tasks. This approach has been evaluated both on a voice command dataset we acquired for the purpose of the study as well as on a publicly available dataset. The experiments show that a single model learned on unaligned data is competitive with state-of-the-art models which depend on aligned data.}},
author = {{Mishakova, Anastasiia and Portet, François and Desot, Thierry and Vacher, Michel}},
booktitle = {{2019 IEEE INTERNATIONAL CONFERENCE ON PERVASIVE COMPUTING AND COMMUNICATIONS WORKSHOPS (PERCOM WORKSHOPS)}},
isbn = {{9781538691519}},
issn = {{2474-2503}},
keywords = {{Natural Language Understanding,Smart Environments,Deep Neural Network,Voice-User Interface,AUDIO}},
language = {{eng}},
location = {{Kyoto, JAPAN}},
pages = {{832--837}},
publisher = {{IEEE}},
title = {{Learning natural language understanding systems from unaligned labels for voice command in smart homes}},
url = {{http://doi.org/10.1109/PERCOMW.2019.8730721}},
year = {{2019}},
}
- Altmetric
- View in Altmetric
- Web of Science
- Times cited: