Exploiting Pseudo Image Captions for Multimodal Summarization

TL;DR

This paper addresses false negatives in vision-language contrastive learning by optimizing mutual information more accurately, leading to improved cross-modal summarization performance.

Contribution

It introduces a novel MI optimization strategy that better handles false negatives in VLP, backed by theoretical analysis and improved experimental results.

Findings

Method outperforms baselines on four cross-modal tasks.

Balances false negative effects effectively.

Theoretically justified MI optimization approach.

Abstract

Cross-modal contrastive learning in vision language pretraining (VLP) faces the challenge of (partial) false negatives. In this paper, we study this problem from the perspective of Mutual Information (MI) optimization. It is common sense that InfoNCE loss used in contrastive learning will maximize the lower bound of MI between anchors and their positives, while we theoretically prove that MI involving negatives also matters when noises commonly exist. Guided by a more general lower bound form for optimization, we propose a contrastive learning strategy regulated by progressively refined cross-modal similarity, to more accurately optimize MI between an image/text anchor and its negative texts/images instead of improperly minimizing it. Our method performs competitively on four downstream cross-modal tasks and systematically balances the beneficial and harmful effects of (partial) false negative samples under theoretical guidance.