Skip to content

Official implementation of Matrix Variational Masked Autoencoder (M-MAE) for paper "Information Flow in Self-Supervised Learning" (https://arxiv.org/abs/2309.17281)

Notifications You must be signed in to change notification settings

MIFA-Lab/M-MAE

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

d4164fc · Oct 6, 2023

History

7 Commits
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Oct 6, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Oct 6, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023
Sep 27, 2023

Repository files navigation

M-MAE (Matrix Variational Masked Autoencoder)

Official implementation of Matrix Variational Masked Autoencoder (M-MAE) for paper "Information Flow in Self-Supervised Learning" (https://arxiv.org/abs/2309.17281).

This repository includes a PyTorch implementation of the Matrix Variational Masked Autoencoder (M-MAE). M-MAE is an extension of MAE (He et al., 2022) and U-MAE (Zhang et al., 2022) by further encouraging the feature uniformity of MAE from a matrix information theoretic perspective.

Instructions

This repo is based on the official code of MAE and official code of U-MAE with minor modifications below, and we follow all the default training and evaluation configurations of MAE. Please see their instructions README_mae.md for details.

Main differences. In M-MAE, we introduce a uniformity_loss_TCR (implemented in loss_func.py) as a uniformity regularization to the MAE loss. It also introduces an additional hyper-parameter lamb (default to 1e-2) in pretrain.sh, which represents the coefficient of the uniformity regularization in the M-MAE loss.

Minor points:

  1. We add a linear classifier to monitor the online linear accuracy and its gradient will not be backpropagated to the backbone encoder.
  2. For efficiency, we only train M-MAE for 200 epochs, and accordingly, we adopt 20 warmup epochs.

Acknowledgement

Our code follows the official implementations of MAE (https://github.com/facebookresearch/mae) and U-MAE (https://github.com/zhangq327/U-MAE). We thank the authors for their great work.

About

Official implementation of Matrix Variational Masked Autoencoder (M-MAE) for paper "Information Flow in Self-Supervised Learning" (https://arxiv.org/abs/2309.17281)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 98.5%
  • Shell 1.5%