Expectation-Maximization Contrastive Learning for Compact Video-and-Language Representations

TL;DR

This paper introduces EMCL, a novel contrastive learning method using expectation-maximization to learn compact, more discriminative video-and-language representations, significantly improving retrieval performance.

Contribution

The paper proposes EMCL, a new approach that finds a compact basis for the latent space, reducing its rank and enhancing semantic representation power in video-and-language tasks.

Findings

Outperforms previous state-of-the-art methods on three benchmark datasets.

Enhances representation discriminability and retrieval accuracy.

Can be integrated into existing models without additional training.

Abstract

Most video-and-language representation learning approaches employ contrastive learning, e.g., CLIP, to project the video and text features into a common latent space according to the semantic similarities of text-video pairs. However, such learned shared latent spaces are not often optimal, and the modality gap between visual and textual representation can not be fully eliminated. In this paper, we propose Expectation-Maximization Contrastive Learning (EMCL) to learn compact video-and-language representations. Specifically, we use the Expectation-Maximization algorithm to find a compact set of bases for the latent space, where the features could be concisely represented as the linear combinations of these bases. Such feature decomposition of video-and-language representations reduces the rank of the latent space, resulting in increased representing power for the semantics. Extensive experiments on three benchmark text-video retrieval datasets prove that our EMCL can learn more discriminative video-and-language representations than previous methods, and significantly outperform previous state-of-the-art methods across all metrics. More encouragingly, the proposed method can be applied to boost the performance of existing approaches either as a jointly training layer or an out-of-the-box inference module with no extra training, making it easy to be incorporated into any existing methods.