TreeMix: Compositional constituency-based data augmentation for natural language understanding
Data augmentation is an effective approach to tackle over-fitting. Many previous works have
proposed different data augmentations strategies for NLP, such as noise injection, word …
proposed different data augmentations strategies for NLP, such as noise injection, word …
Substructure substitution: Structured data augmentation for NLP
We study a family of data augmentation methods, substructure substitution (SUB2), for
natural language processing (NLP) tasks. SUB2 generates new examples by substituting …
natural language processing (NLP) tasks. SUB2 generates new examples by substituting …
Audio-visual neural syntax acquisition
We study phrase structure induction from visually-grounded speech. The core idea is to first
segment the speech waveform into sequences of word segments, and subsequently induce …
segment the speech waveform into sequences of word segments, and subsequently induce …
Unsupervised chunking with hierarchical RNN
In Natural Language Processing (NLP), predicting linguistic structures, such as parsing and
chunking, has mostly relied on manual annotations of syntactic structures. This paper …
chunking, has mostly relied on manual annotations of syntactic structures. This paper …
PCFGs can do better: Inducing probabilistic context-free grammars with many symbols
Probabilistic context-free grammars (PCFGs) with neural parameterization have been shown
to be effective in unsupervised phrase-structure grammar induction. However, due to the …
to be effective in unsupervised phrase-structure grammar induction. However, due to the …
Heads-up! unsupervised constituency parsing via self-attention heads
Transformer-based pre-trained language models (PLMs) have dramatically improved the
state of the art in NLP across many tasks. This has led to substantial interest in analyzing the …
state of the art in NLP across many tasks. This has led to substantial interest in analyzing the …
Data augmentation for machine translation via dependency subtree swapping
We present a generic framework for data augmentation via dependency subtree swapping
that is applicable to machine translation. We extract corresponding subtrees from the …
that is applicable to machine translation. We extract corresponding subtrees from the …
Learning a grammar inducer from massive uncurated instructional videos
Video-aided grammar induction aims to leverage video information for finding more accurate
syntactic grammars for accompanying text. While previous work focuses on building systems …
syntactic grammars for accompanying text. While previous work focuses on building systems …
Revisiting the practical effectiveness of constituency parse extraction from pre-trained language models
T Kim - arXiv preprint arXiv:2211.00479, 2022 - arxiv.org
Constituency Parse Extraction from Pre-trained Language Models (CPE-PLM) is a recent
paradigm that attempts to induce constituency parse trees relying only on the internal …
paradigm that attempts to induce constituency parse trees relying only on the internal …
Unsupervised discontinuous constituency parsing with mildly context-sensitive grammars
We study grammar induction with mildly context-sensitive grammars for unsupervised
discontinuous parsing. Using the probabilistic linear context-free rewriting system (LCFRS) …
discontinuous parsing. Using the probabilistic linear context-free rewriting system (LCFRS) …