Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders

Andrew Drozdov, Pat Verga, Mohit Yadav, Mohit Iyyer, Andrew McCallum
4/3/2019
cs.CL

Official code: iesl/diora (89 stars, Apache-2.0, last commit 2022-01-17).

Abstract

We introduce deep inside-outside recursive autoencoders (DIORA), a fully-unsupervised method for discovering syntax that simultaneously learns representations for constituents within the induced tree. Our approach predicts each word in an input sentence conditioned on the rest of the sentence and uses inside-outside dynamic programming to consider all possible binary trees over the sentence. At test time the CKY algorithm extracts the highest scoring parse. DIORA achieves a new state-of-the-art F1 in unsupervised binary constituency parsing (unlabeled) in two benchmark datasets, WSJ and MultiNLI.

View on arXivView PDF

Code Implementations(1)

iesl/dioraOfficial100%
Apache-2.0

Deep Inside-Outside Recursive Autoencoder

8923C, PerlNov 2, 20184 years agoApache-2.0

Similar Papers

Constituency Parsing with a Self-Attentive Encoder

81% match1 repo9132018

Ensemble Distillation for Unsupervised Constituency Parsing

81% match1 repo52023

Recursive Tree Grammar Autoencoders

80% match2020

Tetra-Tagging: Word-Synchronous Parsing with Linear-Time Inference

80% match1 repo142019

Cascading and Direct Approaches to Unsupervised Constituency Parsing on Spoken Sentences

79% match1 repo32023

Constituency Parsing using LLMs

79% match2023

Cite this paper

@article{drozdov2019unsupervised,
  title  = {Unsupervised Latent Tree Induction with Deep Inside-Outside Recursive Autoencoders},
  author = {Andrew Drozdov and Pat Verga and Mohit Yadav and Mohit Iyyer and Andrew McCallum},
  year   = {2019},
  eprint = {1904.02142},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1904.02142}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1904.02142.svg)](https://codeofpaper.com/paper/1904.02142)

Discussion