Re-evaluating the Need for Multimodal Signals in Unsupervised Grammar Induction

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Boyi, Corona, Rodolfo, Mangalam, Karttikeya, Chen, Catherine, Flaherty, Daniel, Belongie, Serge, Weinberger, Kilian Q., Malik, Jitendra, Darrell, Trevor, Klein, Dan
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929312345620480
author Li, Boyi
Corona, Rodolfo
Mangalam, Karttikeya
Chen, Catherine
Flaherty, Daniel
Belongie, Serge
Weinberger, Kilian Q.
Malik, Jitendra
Darrell, Trevor
Klein, Dan
author_facet Li, Boyi
Corona, Rodolfo
Mangalam, Karttikeya
Chen, Catherine
Flaherty, Daniel
Belongie, Serge
Weinberger, Kilian Q.
Malik, Jitendra
Darrell, Trevor
Klein, Dan
contents Are multimodal inputs necessary for grammar induction? Recent work has shown that multimodal training inputs can improve grammar induction. However, these improvements are based on comparisons to weak text-only baselines that were trained on relatively little textual data. To determine whether multimodal inputs are needed in regimes with large amounts of textual training data, we design a stronger text-only baseline, which we refer to as LC-PCFG. LC-PCFG is a C-PFCG that incorporates em-beddings from text-only large language models (LLMs). We use a fixed grammar family to directly compare LC-PCFG to various multi-modal grammar induction methods. We compare performance on four benchmark datasets. LC-PCFG provides an up to 17% relative improvement in Corpus-F1 compared to state-of-the-art multimodal grammar induction methods. LC-PCFG is also more computationally efficient, providing an up to 85% reduction in parameter count and 8.8x reduction in training time compared to multimodal approaches. These results suggest that multimodal inputs may not be necessary for grammar induction, and emphasize the importance of strong vision-free baselines for evaluating the benefit of multimodal approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2212_10564
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Re-evaluating the Need for Multimodal Signals in Unsupervised Grammar Induction
Li, Boyi
Corona, Rodolfo
Mangalam, Karttikeya
Chen, Catherine
Flaherty, Daniel
Belongie, Serge
Weinberger, Kilian Q.
Malik, Jitendra
Darrell, Trevor
Klein, Dan
Computation and Language
Artificial Intelligence
Machine Learning
Are multimodal inputs necessary for grammar induction? Recent work has shown that multimodal training inputs can improve grammar induction. However, these improvements are based on comparisons to weak text-only baselines that were trained on relatively little textual data. To determine whether multimodal inputs are needed in regimes with large amounts of textual training data, we design a stronger text-only baseline, which we refer to as LC-PCFG. LC-PCFG is a C-PFCG that incorporates em-beddings from text-only large language models (LLMs). We use a fixed grammar family to directly compare LC-PCFG to various multi-modal grammar induction methods. We compare performance on four benchmark datasets. LC-PCFG provides an up to 17% relative improvement in Corpus-F1 compared to state-of-the-art multimodal grammar induction methods. LC-PCFG is also more computationally efficient, providing an up to 85% reduction in parameter count and 8.8x reduction in training time compared to multimodal approaches. These results suggest that multimodal inputs may not be necessary for grammar induction, and emphasize the importance of strong vision-free baselines for evaluating the benefit of multimodal approaches.
title Re-evaluating the Need for Multimodal Signals in Unsupervised Grammar Induction
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2212.10564