Vokenization: Improving Language Understanding with Contextualized, Visual-Grounded Supervision

Hao Tan, Mohit Bansal


Abstract
Humans learn language by listening, speaking, writing, reading, and also, via interaction with the multimodal real world. Existing language pre-training frameworks show the effectiveness of text-only self-supervision while we explore the idea of a visually-supervised language model in this paper. We find that the main reason hindering this exploration is the large divergence in magnitude and distributions between the visually-grounded language datasets and pure-language corpora. Therefore, we develop a technique named “vokenization” that extrapolates multimodal alignments to language-only data by contextually mapping language tokens to their related images (which we call “vokens”). The “vokenizer” is trained on relatively small image captioning datasets and we then apply it to generate vokens for large language corpora. Trained with these contextually generated vokens, our visually-supervised language models show consistent improvements over self-supervised alternatives on multiple pure-language tasks such as GLUE, SQuAD, and SWAG.
Anthology ID:
2020.emnlp-main.162
Volume:
Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Month:
November
Year:
2020
Address:
Online
Editors:
Bonnie Webber, Trevor Cohn, Yulan He, Yang Liu
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2066–2080
Language:
URL:
https://aclanthology.org/2020.emnlp-main.162
DOI:
10.18653/v1/2020.emnlp-main.162
Bibkey:
Cite (ACL):
Hao Tan and Mohit Bansal. 2020. Vokenization: Improving Language Understanding with Contextualized, Visual-Grounded Supervision. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2066–2080, Online. Association for Computational Linguistics.
Cite (Informal):
Vokenization: Improving Language Understanding with Contextualized, Visual-Grounded Supervision (Tan & Bansal, EMNLP 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.emnlp-main.162.pdf
Video:
 https://slideslive.com/38939320
Code
 airsplay/vokenization
Data
Conceptual CaptionsGLUEMS COCOMultiNLIQNLISQuADSSTSST-2SWAGVisual Genome