Multimodal Learning Without Labeled Multimodal Data: Guarantees and Applications

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liang, Paul Pu, Ling, Chun Kai, Cheng, Yun, Obolenskiy, Alex, Liu, Yudong, Pandey, Rohan, Wilf, Alex, Morency, Louis-Philippe, Salakhutdinov, Ruslan
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910483914686464
author Liang, Paul Pu
Ling, Chun Kai
Cheng, Yun
Obolenskiy, Alex
Liu, Yudong
Pandey, Rohan
Wilf, Alex
Morency, Louis-Philippe
Salakhutdinov, Ruslan
author_facet Liang, Paul Pu
Ling, Chun Kai
Cheng, Yun
Obolenskiy, Alex
Liu, Yudong
Pandey, Rohan
Wilf, Alex
Morency, Louis-Philippe
Salakhutdinov, Ruslan
contents In many machine learning systems that jointly learn from multiple modalities, a core research question is to understand the nature of multimodal interactions: how modalities combine to provide new task-relevant information that was not present in either alone. We study this challenge of interaction quantification in a semi-supervised setting with only labeled unimodal data and naturally co-occurring multimodal data (e.g., unlabeled images and captions, video and corresponding audio) but when labeling them is time-consuming. Using a precise information-theoretic definition of interactions, our key contribution is the derivation of lower and upper bounds to quantify the amount of multimodal interactions in this semi-supervised setting. We propose two lower bounds: one based on the shared information between modalities and the other based on disagreement between separately trained unimodal classifiers, and derive an upper bound through connections to approximate algorithms for min-entropy couplings. We validate these estimated bounds and show how they accurately track true interactions. Finally, we show how these theoretical results can be used to estimate multimodal model performance, guide data collection, and select appropriate multimodal models for various tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2306_04539
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Multimodal Learning Without Labeled Multimodal Data: Guarantees and Applications
Liang, Paul Pu
Ling, Chun Kai
Cheng, Yun
Obolenskiy, Alex
Liu, Yudong
Pandey, Rohan
Wilf, Alex
Morency, Louis-Philippe
Salakhutdinov, Ruslan
Machine Learning
Computation and Language
Computer Vision and Pattern Recognition
Information Theory
In many machine learning systems that jointly learn from multiple modalities, a core research question is to understand the nature of multimodal interactions: how modalities combine to provide new task-relevant information that was not present in either alone. We study this challenge of interaction quantification in a semi-supervised setting with only labeled unimodal data and naturally co-occurring multimodal data (e.g., unlabeled images and captions, video and corresponding audio) but when labeling them is time-consuming. Using a precise information-theoretic definition of interactions, our key contribution is the derivation of lower and upper bounds to quantify the amount of multimodal interactions in this semi-supervised setting. We propose two lower bounds: one based on the shared information between modalities and the other based on disagreement between separately trained unimodal classifiers, and derive an upper bound through connections to approximate algorithms for min-entropy couplings. We validate these estimated bounds and show how they accurately track true interactions. Finally, we show how these theoretical results can be used to estimate multimodal model performance, guide data collection, and select appropriate multimodal models for various tasks.
title Multimodal Learning Without Labeled Multimodal Data: Guarantees and Applications
topic Machine Learning
Computation and Language
Computer Vision and Pattern Recognition
Information Theory
url https://arxiv.org/abs/2306.04539