TLDR: Twin Learning for Dimensionality Reduction

TL;DR

TLDR introduces a scalable, self-supervised approach for dimensionality reduction that leverages nearest neighbors and redundancy reduction loss, achieving improved retrieval performance and compression efficiency across image and document datasets.

Contribution

The paper adapts self-supervised learning to dimensionality reduction, providing a simple, scalable method applicable to various representations with broad utility.

Findings

+4% mAP over PCA on ROxford dataset

Improves DINO performance on ImageNet

Retains performance with 10x compression

Abstract

Dimensionality reduction methods are unsupervised approaches which learn low-dimensional spaces where some properties of the initial space, typically the notion of "neighborhood", are preserved. Such methods usually require propagation on large k-NN graphs or complicated optimization solvers. On the other hand, self-supervised learning approaches, typically used to learn representations from scratch, rely on simple and more scalable frameworks for learning. In this paper, we propose TLDR, a dimensionality reduction method for generic input spaces that is porting the recent self-supervised learning framework of Zbontar et al. (2021) to the specific task of dimensionality reduction, over arbitrary representations. We propose to use nearest neighbors to build pairs from a training set and a redundancy reduction loss to learn an encoder that produces representations invariant across such pairs. TLDR is a method that is simple, easy to train, and of broad applicability; it consists of an offline nearest neighbor computation step that can be highly approximated, and a straightforward learning process. Aiming for scalability, we focus on improving linear dimensionality reduction, and show consistent gains on image and document retrieval tasks, e.g. gaining +4% mAP over PCA on ROxford for GeM- AP, improving the performance of DINO on ImageNet or retaining it with a 10x compression.