CSA: Data-efficient Mapping of Unimodal Features to Multimodal Features

Po-han Li, Sandeep P. Chinchali, ufuk topcu

Keywords: multimodal, representation learning, relative representations

TL;DR: Canonical similarity analysis (CSA) achieves CLIP-level performance in multimodal tasks using much less data. It maps unimodal features into a multimodal space without extensive GPU training and supports all modalities.

Abstract:

Multimodal encoders like CLIP excel in tasks such as zero-shot image classification and cross-modal retrieval. However, they require excessive training data.
We propose canonical similarity analysis (CSA), which uses two unimodal encoders to replicate multimodal encoders using limited data.
CSA maps unimodal features into a multimodal space, using a new similarity score to retain only the multimodal information.
CSA only involves the inference of unimodal encoders and a cubic-complexity matrix decomposition, eliminating the need for extensive GPU-based model training.
Experiments show that CSA outperforms CLIP while requiring $50,000\times$ fewer multimodal data pairs to bridge the modalities given pre-trained unimodal encoders on ImageNet classification and misinformative news caption detection.
CSA surpasses the state-of-the-art method to map unimodal features to multimodal features.
We also demonstrate the ability of CSA with modalities beyond image and text, paving the way for future modality pairs with limited paired multimodal data but abundant unpaired unimodal data, such as lidar and text.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 2221