Postdoc, Genentech
2 papers at NeurIPS 2025
For datasets with high-magnitude noise features, joint-embedding is more robust than reconstruction for self-supervised learning.
We show that self-supervised models like DINOv2 can develop strong noise robustness without any explicit denoiser at downstream fine-tuning or inference, by leveraging a data curriculum and a denoised regularization loss during pretraining.
