Skin lesion classification datasets often suffer from severe class imbalance, with malignant cases significantly underrepresented, leading to biased decision boundaries during deep learning training. We address this challenge using class-conditioned diffusion models to generate synthetic dermatological images, followed by self-supervised MAE pretraining to enable huge ViT models to learn robust, domain-relevant features. To support deployment in practical clinical settings, where lightweight models are required, we apply knowledge distillation to transfer these representations to a smaller ViT student suitable for mobile devices. Our results show that MAE pretraining on synthetic data, combined with distillation, improves classification performance while enabling efficient on-device inference for practical clinical use.
@article{arxiv.2602.19848,
title = {DerMAE: Improving skin lesion classification through conditioned latent diffusion and MAE distillation},
author = {Francisco Filho and Kelvin Cunha and Fábio Papais and Emanoel dos Santos and Rodrigo Mota and Thales Bezerra and Erico Medeiros and Paulo Borba and Tsang Ing Ren},
journal= {arXiv preprint arXiv:2602.19848},
year = {2026}
}