Some improvements of the spectral learning approach for probabilistic grammatical inference

Mattias Gybels, François Denis, Amaury Habrard
The 12th International Conference on Grammatical Inference, PMLR 34:64-78, 2014.

Abstract

Spectral methods propose new and elegant solutions in probabilistic grammatical inference. We propose two ways to improve them. We show how a linear representation, or equivalently a weighted automata, output by the spectral learning algorithm can be taken as an initial point for the Baum Welch algorithm, in order to increase the likelihood of the observation data. Secondly, we show how the inference problem can naturally be expressed in the framework of Structured Low-Rank Approximation. Both ideas are tested on a benchmark extracted from the PAutomaC challenge.

Cite this Paper

BibTeX
@InProceedings{pmlr-v34-gybels14a, title = {Some improvements of the spectral learning approach for probabilistic grammatical inference}, author = {Gybels, Mattias and Denis, François and Habrard, Amaury}, booktitle = {The 12th International Conference on Grammatical Inference}, pages = {64--78}, year = {2014}, editor = {Clark, Alexander and Kanazawa, Makoto and Yoshinaka, Ryo}, volume = {34}, series = {Proceedings of Machine Learning Research}, address = {Kyoto, Japan}, month = {17--19 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v34/gybels14a.pdf}, url = {https://proceedings.mlr.press/v34/gybels14a.html}, abstract = {Spectral methods propose new and elegant solutions in probabilistic grammatical inference. We propose two ways to improve them. We show how a linear representation, or equivalently a weighted automata, output by the spectral learning algorithm can be taken as an initial point for the Baum Welch algorithm, in order to increase the likelihood of the observation data. Secondly, we show how the inference problem can naturally be expressed in the framework of Structured Low-Rank Approximation. Both ideas are tested on a benchmark extracted from the PAutomaC challenge.} }
Endnote
%0 Conference Paper %T Some improvements of the spectral learning approach for probabilistic grammatical inference %A Mattias Gybels %A François Denis %A Amaury Habrard %B The 12th International Conference on Grammatical Inference %C Proceedings of Machine Learning Research %D 2014 %E Alexander Clark %E Makoto Kanazawa %E Ryo Yoshinaka %F pmlr-v34-gybels14a %I PMLR %P 64--78 %U https://proceedings.mlr.press/v34/gybels14a.html %V 34 %X Spectral methods propose new and elegant solutions in probabilistic grammatical inference. We propose two ways to improve them. We show how a linear representation, or equivalently a weighted automata, output by the spectral learning algorithm can be taken as an initial point for the Baum Welch algorithm, in order to increase the likelihood of the observation data. Secondly, we show how the inference problem can naturally be expressed in the framework of Structured Low-Rank Approximation. Both ideas are tested on a benchmark extracted from the PAutomaC challenge.
RIS
TY - CPAPER TI - Some improvements of the spectral learning approach for probabilistic grammatical inference AU - Mattias Gybels AU - François Denis AU - Amaury Habrard BT - The 12th International Conference on Grammatical Inference DA - 2014/08/30 ED - Alexander Clark ED - Makoto Kanazawa ED - Ryo Yoshinaka ID - pmlr-v34-gybels14a PB - PMLR DP - Proceedings of Machine Learning Research VL - 34 SP - 64 EP - 78 L1 - http://proceedings.mlr.press/v34/gybels14a.pdf UR - https://proceedings.mlr.press/v34/gybels14a.html AB - Spectral methods propose new and elegant solutions in probabilistic grammatical inference. We propose two ways to improve them. We show how a linear representation, or equivalently a weighted automata, output by the spectral learning algorithm can be taken as an initial point for the Baum Welch algorithm, in order to increase the likelihood of the observation data. Secondly, we show how the inference problem can naturally be expressed in the framework of Structured Low-Rank Approximation. Both ideas are tested on a benchmark extracted from the PAutomaC challenge. ER -
APA
Gybels, M., Denis, F. & Habrard, A.. (2014). Some improvements of the spectral learning approach for probabilistic grammatical inference. The 12th International Conference on Grammatical Inference, in Proceedings of Machine Learning Research 34:64-78 Available from https://proceedings.mlr.press/v34/gybels14a.html.

Related Material