Feature Template-Based Parallel Computation Technique for Conditional Random Fields
Abstract: Conditional Random Fields (CRFs) is a popular probabilistic graphical model, which has been applied in a wide range of areas, including Natural Language Process (NLP), Bioinformatics, Computer Vision, etc. However, con- textual features-based methods usually lead to large-scale feature functions and result in high computational complexity and low model training efficiency. In this paper, a feature template-based parallel computation technique is proposed to parallelly process M matrix and reduce computational complexity through observing the main feature of contextual feature function created by the template. Experimental results show that our approach significantly outperforms tradi- tional feature function approach on computation speed.
文章引用: 黄双萍 , 苏志良 , 岳学军 , 邓小玲 (2013) 基于特征模板的条件随机场快速并行计算技术。 计算机科学与应用， 3， 251-256. doi: 10.12677/CSA.2013.35043
 J. Lafferty, A. McCallum and F. Pereira. Conditional random fields: Probabilistic models for segmenting and labeling se-quence data. Proceedings of the Eighteenth International Conference on Machine Learning, 2001: 282-289.
 F. Sha, F. Pereira. Shallow parsing with conditional random fields. Proceedings of the 2003 Conference of the North American Chapter of the Association for Computational Linguistics on Human Language Technology, 2003, (1): 134-141. http://www.cnts.ua.ac.be/conll2000/
 C. M. Bishop. Pattern recognition and machine learning. New York: Springer, 2006: 359-418.
 A. McCallum, D. Freitag and F. Pereira. Maximum entropy markov models, 2000.
 TakuKudo, CRF++ toolkit, 2005. http://crfpp.sourceforge.net/
 S. Della Pietra, V. Della Pietra and J. Lafferty. Inducing features of random fields. IEEE Transactions on Pattern Analysis and Machine In-telligence, 1997, 19(4): 380-393.
 F. Erik, T. K. Sang and S. Buchholz. Introduction to the CoNLL- 2000 Shared Task: Chunking. Proceedings of CoNLL-2000 and LLL-2000, 2000.