Align before Fuse: Vision and Language Representation Learning with Momentum Distillation

TL;DR

This paper introduces ALBEF, a vision-language model that aligns image and text representations before fusion using contrastive learning, and employs momentum distillation to enhance learning from noisy web data, achieving state-of-the-art results.

Contribution

The paper proposes a novel alignment-before-fusion approach with contrastive loss and introduces momentum distillation for improved learning from web data, without requiring bounding boxes.

Findings

ALBEF outperforms larger dataset pre-trained models on image-text retrieval.

ALBEF achieves significant improvements on VQA and NLVR$^2$ tasks.

Theoretical analysis links training tasks to mutual information maximization.

Abstract

Large-scale vision and language representation learning has shown promising improvements on various vision-language tasks. Most existing methods employ a transformer-based multimodal encoder to jointly model visual tokens (region-based image features) and word tokens. Because the visual tokens and word tokens are unaligned, it is challenging for the multimodal encoder to learn image-text interactions. In this paper, we introduce a contrastive loss to ALign the image and text representations BEfore Fusing (ALBEF) them through cross-modal attention, which enables more grounded vision and language representation learning. Unlike most existing methods, our method does not require bounding box annotations nor high-resolution images. In order to improve learning from noisy web data, we propose momentum distillation, a self-training method which learns from pseudo-targets produced by a momentum model. We provide a theoretical analysis of ALBEF from a mutual information maximization perspective, showing that different training tasks can be interpreted as different ways to generate views for an image-text pair. ALBEF achieves state-of-the-art performance on multiple downstream vision-language tasks. On image-text retrieval, ALBEF outperforms methods that are pre-trained on orders of magnitude larger datasets. On VQA and NLVR$^2$, ALBEF achieves absolute improvements of 2.37% and 3.84% compared to the state-of-the-art, while enjoying faster inference speed. Code and pre-trained models are available at https://github.com/salesforce/ALBEF/.