Combining Generative and Discriminative Approaches to Unsupervised Dependency Parsing via Dual Decomposition

Unsupervised dependency parsing aims to learn a dependency parser from unannotated sentences. Existing work focuses on either learning generative models using the expectation-maximization algorithm and its variants, or learning discriminative models using the discriminative clustering algorithm. In this paper, we propose a new learning strategy that learns a generative model and a discriminative model jointly based on the dual decomposition method. Our method is simple and general, yet effective to capture the advantages of both models and improve their learning results. We tested our method on the UD treebank and achieved a state-of-the-art performance on thirty languages.

[1]  Koby Crammer,et al.  Online Large-Margin Training of Dependency Parsers , 2005, ACL.

[2]  Mark Johnson,et al.  Using Left-corner Parsing to Encode Universal Structural Constraints in Grammar Induction , 2016, EMNLP.

[3]  Alexander M. Rush,et al.  A Tutorial on Dual Decomposition and Lagrangian Relaxation for Inference in Natural Language Processing , 2012, J. Artif. Intell. Res..

[4]  Dan Klein,et al.  Corpus-Based Induction of Syntactic Structure: Models of Dependency and Constituency , 2004, ACL.

[5]  Alexander M. Rush,et al.  On Dual Decomposition and Linear Programming Relaxations for Natural Language Processing , 2010, EMNLP.

[6]  Noémie Elhadad,et al.  A convex and feature-rich discriminative approach to dependency grammar induction , 2015, ACL.

[7]  Dan Klein,et al.  Agreement-Based Learning , 2007, NIPS.

[8]  Noah A. Smith,et al.  Shared Logistic Normal Distributions for Soft Parameter Tying in Unsupervised Grammar Induction , 2009, NAACL.

[9]  Alexander M. Rush,et al.  Dual Decomposition for Parsing with Non-Projective Head Automata , 2010, EMNLP.

[10]  Bowen Zhou,et al.  Dependency-based Convolutional Neural Networks for Sentence Embedding , 2015, ACL.

[11]  Yonatan Bisk,et al.  Simple Robust Grammar Induction with Combinatory Categorial Grammars , 2012, AAAI.

[12]  Nikos Komodakis,et al.  MRF Optimization via Dual Decomposition: Message-Passing Revisited , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[13]  Kewei Tu,et al.  Unsupervised Neural Dependency Parsing , 2016, EMNLP.

[14]  Alessandro Moschitti,et al.  High-Order Low-Rank Tensors for Semantic Role Labeling , 2015, HLT-NAACL.

[15]  Noah A. Smith,et al.  Annealing Structural Bias in Multilingual Weighted Grammar Induction , 2006, ACL.

[16]  Noah A. Smith,et al.  Logistic Normal Priors for Unsupervised Probabilistic Grammar Induction , 2008, NIPS.

[17]  George B. Dantzig,et al.  Decomposition Principle for Linear Programs , 1960 .

[18]  John DeNero,et al.  Painless Unsupervised Learning with Features , 2010, NAACL.