Learning Shared Representations from Unpaired Data

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yacobi, Amitai, Ben-Ari, Nir, Talmon, Ronen, Shaham, Uri
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915547800666112
author Yacobi, Amitai
Ben-Ari, Nir
Talmon, Ronen
Shaham, Uri
author_facet Yacobi, Amitai
Ben-Ari, Nir
Talmon, Ronen
Shaham, Uri
contents Learning shared representations is a primary area of multimodal representation learning. The current approaches to achieve a shared embedding space rely heavily on paired samples from each modality, which are significantly harder to obtain than unpaired ones. In this work, we demonstrate that shared representations can be learned almost exclusively from unpaired data. Our arguments are grounded in the spectral embeddings of the random walk matrices constructed independently from each unimodal representation. Empirical results in computer vision and natural language processing domains support its potential, revealing the effectiveness of unpaired data in capturing meaningful cross-modal relations, demonstrating high capabilities in retrieval tasks, generation, arithmetics, zero-shot, and cross-domain classification. This work, to the best of our knowledge, is the first to demonstrate these capabilities almost exclusively from unpaired samples, giving rise to a cross-modal embedding that could be viewed as universal, i.e., independent of the specific modalities of the data. Our project page: https://shaham-lab.github.io/SUE_page.
format Preprint
id arxiv_https___arxiv_org_abs_2505_21524
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning Shared Representations from Unpaired Data
Yacobi, Amitai
Ben-Ari, Nir
Talmon, Ronen
Shaham, Uri
Computer Vision and Pattern Recognition
Machine Learning
Learning shared representations is a primary area of multimodal representation learning. The current approaches to achieve a shared embedding space rely heavily on paired samples from each modality, which are significantly harder to obtain than unpaired ones. In this work, we demonstrate that shared representations can be learned almost exclusively from unpaired data. Our arguments are grounded in the spectral embeddings of the random walk matrices constructed independently from each unimodal representation. Empirical results in computer vision and natural language processing domains support its potential, revealing the effectiveness of unpaired data in capturing meaningful cross-modal relations, demonstrating high capabilities in retrieval tasks, generation, arithmetics, zero-shot, and cross-domain classification. This work, to the best of our knowledge, is the first to demonstrate these capabilities almost exclusively from unpaired samples, giving rise to a cross-modal embedding that could be viewed as universal, i.e., independent of the specific modalities of the data. Our project page: https://shaham-lab.github.io/SUE_page.
title Learning Shared Representations from Unpaired Data
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2505.21524