Matrix Information Theory for Self-Supervised Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yifan, Tan, Zhiquan, Yang, Jingqin, Huang, Weiran, Yuan, Yang
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909315489595392
author Zhang, Yifan
Tan, Zhiquan
Yang, Jingqin
Huang, Weiran
Yuan, Yang
author_facet Zhang, Yifan
Tan, Zhiquan
Yang, Jingqin
Huang, Weiran
Yuan, Yang
contents The maximum entropy encoding framework provides a unified perspective for many non-contrastive learning methods like SimSiam, Barlow Twins, and MEC. Inspired by this framework, we introduce Matrix-SSL, a novel approach that leverages matrix information theory to interpret the maximum entropy encoding loss as matrix uniformity loss. Furthermore, Matrix-SSL enhances the maximum entropy encoding method by seamlessly incorporating matrix alignment loss, directly aligning covariance matrices in different branches. Experimental results reveal that Matrix-SSL outperforms state-of-the-art methods on the ImageNet dataset under linear evaluation settings and on MS-COCO for transfer learning tasks. Specifically, when performing transfer learning tasks on MS-COCO, our method outperforms previous SOTA methods such as MoCo v2 and BYOL up to 3.3% with only 400 epochs compared to 800 epochs pre-training. We also try to introduce representation learning into the language modeling regime by fine-tuning a 7B model using matrix cross-entropy loss, with a margin of 3.1% on the GSM8K dataset over the standard cross-entropy loss. Code available at https://github.com/yifanzhang-pro/Matrix-SSL.
format Preprint
id arxiv_https___arxiv_org_abs_2305_17326
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Matrix Information Theory for Self-Supervised Learning
Zhang, Yifan
Tan, Zhiquan
Yang, Jingqin
Huang, Weiran
Yuan, Yang
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
The maximum entropy encoding framework provides a unified perspective for many non-contrastive learning methods like SimSiam, Barlow Twins, and MEC. Inspired by this framework, we introduce Matrix-SSL, a novel approach that leverages matrix information theory to interpret the maximum entropy encoding loss as matrix uniformity loss. Furthermore, Matrix-SSL enhances the maximum entropy encoding method by seamlessly incorporating matrix alignment loss, directly aligning covariance matrices in different branches. Experimental results reveal that Matrix-SSL outperforms state-of-the-art methods on the ImageNet dataset under linear evaluation settings and on MS-COCO for transfer learning tasks. Specifically, when performing transfer learning tasks on MS-COCO, our method outperforms previous SOTA methods such as MoCo v2 and BYOL up to 3.3% with only 400 epochs compared to 800 epochs pre-training. We also try to introduce representation learning into the language modeling regime by fine-tuning a 7B model using matrix cross-entropy loss, with a margin of 3.1% on the GSM8K dataset over the standard cross-entropy loss. Code available at https://github.com/yifanzhang-pro/Matrix-SSL.
title Matrix Information Theory for Self-Supervised Learning
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2305.17326