Inferring biological relationships from cellular phenotypes in high-content microscopy screens provides significant opportunity and challenge in biological research. Prior results have shown that deep vision models can capture biological signal better than hand-crafted features. This work explores how self-supervised deep learning approaches scale when training larger models on larger microscopy datasets. Our results show that both CNN- and ViT-based masked autoencoders significantly outperform weakly supervised baselines. At the high-end of our scale, a ViT-L/8 trained on over 3.5-billion unique crops sampled from 93-million microscopy images achieves relative improvements as high as 28% over our best weakly supervised baseline at inferring known biological relationships curated from public databases. Relevant code and select models released with this work can be found at: https://github.com/recursionpharma/maes_microscopy.
@article{arxiv.2309.16064,
title = {Masked Autoencoders are Scalable Learners of Cellular Morphology},
author = {Oren Kraus and Kian Kenyon-Dean and Saber Saberian and Maryam Fallah and Peter McLean and Jess Leung and Vasudev Sharma and Ayla Khan and Jia Balakrishnan and Safiye Celik and Maciej Sypetkowski and Chi Vicky Cheng and Kristen Morse and Maureen Makes and Ben Mabey and Berton Earnshaw},
journal= {arXiv preprint arXiv:2309.16064},
year = {2023}
}
Comments
Spotlight at NeurIPS 2023 Generative AI and Biology (GenBio) Workshop