Eigenfunction Extraction for Ordered Representation Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Varıcı, Burak, Tsai, Che-Ping, Ray, Ritabrata, Boffi, Nicholas M., Ravikumar, Pradeep
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915583337955328
author Varıcı, Burak
Tsai, Che-Ping
Ray, Ritabrata
Boffi, Nicholas M.
Ravikumar, Pradeep
author_facet Varıcı, Burak
Tsai, Che-Ping
Ray, Ritabrata
Boffi, Nicholas M.
Ravikumar, Pradeep
contents Recent advances in representation learning reveal that widely used objectives, such as contrastive and non-contrastive, implicitly perform spectral decomposition of a contextual kernel, induced by the relationship between inputs and their contexts. Yet, these methods recover only the linear span of top eigenfunctions of the kernel, whereas exact spectral decomposition is essential for understanding feature ordering and importance. In this work, we propose a general framework to extract ordered and identifiable eigenfunctions, based on modular building blocks designed to satisfy key desiderata, including compatibility with the contextual kernel and scalability to modern settings. We then show how two main methodological paradigms, low-rank approximation and Rayleigh quotient optimization, align with this framework for eigenfunction extraction. Finally, we validate our approach on synthetic kernels and demonstrate on real-world image datasets that the recovered eigenvalues act as effective importance scores for feature selection, enabling principled efficiency-accuracy tradeoffs via adaptive-dimensional representations.
format Preprint
id arxiv_https___arxiv_org_abs_2510_24672
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Eigenfunction Extraction for Ordered Representation Learning
Varıcı, Burak
Tsai, Che-Ping
Ray, Ritabrata
Boffi, Nicholas M.
Ravikumar, Pradeep
Machine Learning
Recent advances in representation learning reveal that widely used objectives, such as contrastive and non-contrastive, implicitly perform spectral decomposition of a contextual kernel, induced by the relationship between inputs and their contexts. Yet, these methods recover only the linear span of top eigenfunctions of the kernel, whereas exact spectral decomposition is essential for understanding feature ordering and importance. In this work, we propose a general framework to extract ordered and identifiable eigenfunctions, based on modular building blocks designed to satisfy key desiderata, including compatibility with the contextual kernel and scalability to modern settings. We then show how two main methodological paradigms, low-rank approximation and Rayleigh quotient optimization, align with this framework for eigenfunction extraction. Finally, we validate our approach on synthetic kernels and demonstrate on real-world image datasets that the recovered eigenvalues act as effective importance scores for feature selection, enabling principled efficiency-accuracy tradeoffs via adaptive-dimensional representations.
title Eigenfunction Extraction for Ordered Representation Learning
topic Machine Learning
url https://arxiv.org/abs/2510.24672