Learning Kernels over Strings using Gaussian Processes

Daniel Beck, Trevor Cohn


Abstract
Non-contiguous word sequences are widely known to be important in modelling natural language. However they not explicitly encoded in common text representations. In this work we propose a model for text processing using string kernels, capable of flexibly representing non-contiguous sequences. Specifically, we derive a vectorised version of the string kernel algorithm and their gradients, allowing efficient hyperparameter optimisation as part of a Gaussian Process framework. Experiments on synthetic data and text regression for emotion analysis show the promise of this technique.
Anthology ID:
I17-2012
Volume:
Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers)
Month:
November
Year:
2017
Address:
Taipei, Taiwan
Editors:
Greg Kondrak, Taro Watanabe
Venue:
IJCNLP
SIG:
Publisher:
Asian Federation of Natural Language Processing
Note:
Pages:
67–73
Language:
URL:
https://aclanthology.org/I17-2012
DOI:
Bibkey:
Cite (ACL):
Daniel Beck and Trevor Cohn. 2017. Learning Kernels over Strings using Gaussian Processes. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 67–73, Taipei, Taiwan. Asian Federation of Natural Language Processing.
Cite (Informal):
Learning Kernels over Strings using Gaussian Processes (Beck & Cohn, IJCNLP 2017)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-1/I17-2012.pdf
Note:
 I17-2012.Notes.pdf