> ML_LITERATURE // HE-2022-MASKED-AUTOENCODERS-SCALABLE-VISION-LEARNERS-MAE_v1.0
Masked Autoencoders Are Scalable Vision Learners (MAE)
Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, Ross Girshick · IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2022)
algorithm2022foundationalthirdPartyReproduced
Principal Contribution
Masked 75% of random image patches and applied an asymmetric Vision Transformer where the heavy encoder processes only visible patches, accelerating pre-training by 3x-4x.
Operational Relevance
Serves as qualified reference for deploying task-representation-learning, task-image-classification in production.
Assumptions
- Spatial feature coherence and data manifold structure adhere to continuous representation hypotheses
Limitations
- Computational complexity scales with spatial resolution and parameter capacity