URLOST: Unsupervised Representation Learning without Stationarity or Topology

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Yun, Zeyu, Zhang, Juexiao, LeCun, Yann, Chen, Yubei
Format: Preprint
Publié: 2023
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866912284388884480
author Yun, Zeyu
Zhang, Juexiao
LeCun, Yann
Chen, Yubei
author_facet Yun, Zeyu
Zhang, Juexiao
LeCun, Yann
Chen, Yubei
contents Unsupervised representation learning has seen tremendous progress. However, it is constrained by its reliance on domain specific stationarity and topology, a limitation not found in biological intelligence systems. For instance, unlike computer vision, human vision can process visual signals sampled from highly irregular and non-stationary sensors. We introduce a novel framework that learns from high-dimensional data without prior knowledge of stationarity and topology. Our model, abbreviated as URLOST, combines a learnable self-organizing layer, spectral clustering, and a masked autoencoder (MAE). We evaluate its effectiveness on three diverse data modalities including simulated biological vision data, neural recordings from the primary visual cortex, and gene expressions. Compared to state-of-the-art unsupervised learning methods like SimCLR and MAE, our model excels at learning meaningful representations across diverse modalities without knowing their stationarity or topology. It also outperforms other methods that are not dependent on these factors, setting a new benchmark in the field. We position this work as a step toward unsupervised learning methods capable of generalizing across diverse high-dimensional data modalities.
format Preprint
id arxiv_https___arxiv_org_abs_2310_04496
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle URLOST: Unsupervised Representation Learning without Stationarity or Topology
Yun, Zeyu
Zhang, Juexiao
LeCun, Yann
Chen, Yubei
Computer Vision and Pattern Recognition
Machine Learning
Unsupervised representation learning has seen tremendous progress. However, it is constrained by its reliance on domain specific stationarity and topology, a limitation not found in biological intelligence systems. For instance, unlike computer vision, human vision can process visual signals sampled from highly irregular and non-stationary sensors. We introduce a novel framework that learns from high-dimensional data without prior knowledge of stationarity and topology. Our model, abbreviated as URLOST, combines a learnable self-organizing layer, spectral clustering, and a masked autoencoder (MAE). We evaluate its effectiveness on three diverse data modalities including simulated biological vision data, neural recordings from the primary visual cortex, and gene expressions. Compared to state-of-the-art unsupervised learning methods like SimCLR and MAE, our model excels at learning meaningful representations across diverse modalities without knowing their stationarity or topology. It also outperforms other methods that are not dependent on these factors, setting a new benchmark in the field. We position this work as a step toward unsupervised learning methods capable of generalizing across diverse high-dimensional data modalities.
title URLOST: Unsupervised Representation Learning without Stationarity or Topology
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2310.04496