Dynamic Feature Induction: The Last Gist to the State-of-the-Art
作者： Jinho D. Choi
DOI: 10.18653/V1/N16-1031
关键词:
摘要: We introduce a novel technique called dynamic feature induction that keeps inducing high dimensional features automatically until the space becomes ‘more’ linearly separable. Dynamic searches for combinations give strong clues distinguishing certain label pairs, and generates joint from these combinations. These induced are trained along with primitive low features. Our approach was evaluated on two core NLP tasks, part-of-speech tagging named entity recognition, showed state-of-the-art results both achieving accuracy of 97.64 F1-score 91.00 respectively, about 25% increase in space.
aclweb.org参考文章(49)
Nianwen Xue, Martha Palmer, None, Calibrating Features for Semantic Role Labeling empirical methods in natural language processing. pp. 88- 94 ,(2004)
Joakim Nivre, Yoav Goldberg, A Dynamic Oracle for Arc-Eager Dependency Parsing international conference on computational linguistics. pp. 959- 976 ,(2012)
Christopher D. Manning, Part-of-speech tagging from 97% to 100%: is it time for some linguistics? international conference on computational linguistics. pp. 171- 189 ,(2011) , 10.1007/978-3-642-19400-9_14
Behzad Mortazavi-Asl, Umeshwar Dayal, Qiming Chen, Jiawei Han, Jian Pei, Meichun Hsu, Helen Pinto, PrefixSpan: Mining Sequential Patterns by Prefix-Projected Growth international conference on data engineering. pp. 215- 224 ,(2001)
Phil Blunsom, Stephen Pulman, Karl Moritz Hermann, Lei Yu, Deep learning for answer sentence selection arXiv: Computation and Language. ,(2014)
Mark Dredze, Kuzman Ganchev, Small Statistical Models by Random Feature Mixing meeting of the association for computational linguistics. pp. 19- 20 ,(2008)
Mitch Marcus, Beatrice Santorini, Mary Ann Marcinkiewicz, None, Building a large annotated corpus of English: the penn treebank Computational Linguistics. ,vol. 19, pp. 313- 330 ,(1993) , 10.21236/ADA273556
James Theiler, Simon Perkins, Kevin Lacker, Grafting: fast, incremental feature selection by gradient descent in function space Journal of Machine Learning Research. ,vol. 3, pp. 1333- 1356 ,(2003)
Hal Daume, Kai-Wei Chang, Akshay Krishnamurthy, Alekh Agarwal, John Langford, Learning to Search Better than Your Teacher international conference on machine learning. pp. 2058- 2066 ,(2015)
Kristina Toutanova, Dan Klein, Christopher D. Manning, Yoram Singer, Feature-rich part-of-speech tagging with a cyclic dependency network Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology - NAACL '03. pp. 173- 180 ,(2003) , 10.3115/1073445.1073478