Modeling Label Correlations for Second-Order Semantic Dependency Parsing with Mean-Field Inference
Not on the map. Why?
Abstract (arXiv)
Second-order semantic parsing with end-to-end mean-field inference has been shown good performance. In this work we aim to improve this method by modeling label correlations between adjacent arcs. However, direct modeling leads to memory explosion because second-order score tensors have sizes of O(n3L2) (n is the sentence length and L is the number of labels), which is not affordable. To tackle this computational challenge, we leverage tensor decomposition techniques, and interestingly, we show that the large second-order score tensors have no need to be materialized during mean-field inference, thereby reducing the computational complexity from cubic to quadratic. We conduct experiments on SemEval 2015 Task 18 English datasets, showing the effectiveness of modeling label correlations. Our code is publicly available at https://github.com/sustcsonglin/mean-field-dep-parsing.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 25 of its 63 references that are in the map. First posted on arXiv 7 Apr 2022.
Builds on
25 of 63 references.
References · 25
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
- 2016Deep Biaffine Attention for Neural Dependency Parsingcited by 334
- 2015Conditional Random Fields as Recurrent Neural Networkscited by 214
- 2020Efficient Second-Order TreeCRF for Neural Dependency Parsingcited by 48
- 2017Introduction to Tensor Decompositions and their Applications in Machine Learningcited by 46
- 2018Simpler but More Accurate Semantic Dependency Parsingcited by 46
- 2019Broad-Coverage Semantic Parsing as Transductioncited by 32
- 2021PCFGs Can Do Better: Inducing Probabilistic Context-Free Grammars with Many Symbolscited by 22
- 2019Left-to-Right Dependency Parsing with Pointer Networkscited by 21
- 2019Second-Order Semantic Dependency Parsing with End-to-End Neural Networkscited by 21
- 2019Establishing Strong Baselines for the New Decade: Sequence Tagging, Syntactic and Semantic Parsing with BERTcited by 19
- 2021Neural Bi-Lexicalized PCFG Inductioncited by 18
13 more
- 2020LP-SparseMAP: Differentiable Relaxed Optimization for Sparse Structured Predictioncited by 17
- 2020Second-Order Neural Dependency Parsing with Message Passing and End-to-End Trainingcited by 17
- 2020Transition-based Semantic Dependency Parsing with Pointer Networkscited by 16
- 2019Compositional Semantic Parsing Across Graphbankscited by 15
- 2020High-order Semantic Role Labelingcited by 11
- 2017Deep Multitask Learning for Semantic Dependency Parsingcited by 10
- 2020Fast semantic parsing with well-typedness guaranteescited by 10
- 2021Regularized Frank-Wolfe for Dense CRFs: Generalizing Mean Field and Beyondcited by 10
- 2019Multi-Task Semantic Dependency Parsing with Policy Gradient for Learning Easy-First Strategiescited by 9
- 2014Conditional Random Field Autoencoders for Unsupervised Structured Predictioncited by 8
- 2015Approximation-Aware Dependency Parsing by Belief Propagationcited by 6
- 2020Neuralizing Efficient Higher-order Belief Propagationcited by 5
- 2021Sparse Fuzzy Attention for Structured Sentiment Analysiscited by 5
Built on by
Citing papers per year: 2022: 1.
1 paper.