Recent advancements in self-supervised learning have reduced the gap between supervised and unsupervised representation learning. However, most self-supervised and deep clustering techniques rely heavily on data augmentation, rendering them ineffective for many learning tasks where insufficient domain knowledge exists for performing augmentation. We propose a new self-distillation based algorithm for domain-agnostic clustering. Our method builds upon the existing deep clustering frameworks and requires no separate student model. The proposed method outperforms existing domain agnostic (augmentation-free) algorithms on CIFAR-10. We empirically demonstrate that knowledge distillation can improve unsupervised representation learning by extracting richer `dark knowledge' from the model than using predicted labels alone. Preliminary experiments also suggest that self-distillation improves the convergence of DeepCluster-v2.
@article{arxiv.2111.12170,
title = {Domain-Agnostic Clustering with Self-Distillation},
author = {Mohammed Adnan and Yani A. Ioannou and Chuan-Yung Tsai and Graham W. Taylor},
journal= {arXiv preprint arXiv:2111.12170},
year = {2021}
}
Comments
NeurIPS 2021 Workshop: Self-Supervised Learning - Theory and Practice