A Survey on Masked Autoencoder for Self-supervised Learning in Vision and Beyond

TL;DR

This paper provides a comprehensive survey of masked autoencoders, highlighting their role in self-supervised learning for vision and their potential to bridge the gap with NLP methods like BERT.

Contribution

It is the first survey to review SSL with masked autoencoders in vision, covering historical development, recent progress, and future implications.

Findings

Masked autoencoders have revived interest in generative SSL in vision.

They show promise in bridging vision and NLP SSL techniques.

The survey discusses diverse applications and future directions.

Abstract

Masked autoencoders are scalable vision learners, as the title of MAE \cite{he2022masked}, which suggests that self-supervised learning (SSL) in vision might undertake a similar trajectory as in NLP. Specifically, generative pretext tasks with the masked prediction (e.g., BERT) have become a de facto standard SSL practice in NLP. By contrast, early attempts at generative methods in vision have been buried by their discriminative counterparts (like contrastive learning); however, the success of mask image modeling has revived the masking autoencoder (often termed denoising autoencoder in the past). As a milestone to bridge the gap with BERT in NLP, masked autoencoder has attracted unprecedented attention for SSL in vision and beyond. This work conducts a comprehensive survey of masked autoencoders to shed insight on a promising direction of SSL. As the first to review SSL with masked autoencoders, this work focuses on its application in vision by discussing its historical developments, recent progress, and implications for diverse applications.