On the Efficacy of Small Self-Supervised Contrastive Models without Distillation Signals
This work addresses the challenge of deploying self-supervised learning in resource-restricted scenarios by enabling effective training of small models without costly distillation, which is incremental as it builds on existing techniques to solve a specific bottleneck.
The paper tackles the problem of training small self-supervised contrastive models without using distillation from large models, showing that small models can complete pretext tasks without overfitting but suffer from over-clustering, and by combining validated techniques, it improves baseline performances of five small architectures with considerable margins.
It is a consensus that small models perform quite poorly under the paradigm of self-supervised contrastive learning. Existing methods usually adopt a large off-the-shelf model to transfer knowledge to the small one via distillation. Despite their effectiveness, distillation-based methods may not be suitable for some resource-restricted scenarios due to the huge computational expenses of deploying a large model. In this paper, we study the issue of training self-supervised small models without distillation signals. We first evaluate the representation spaces of the small models and make two non-negligible observations: (i) the small models can complete the pretext task without overfitting despite their limited capacity and (ii) they universally suffer the problem of over clustering. Then we verify multiple assumptions that are considered to alleviate the over-clustering phenomenon. Finally, we combine the validated techniques and improve the baseline performances of five small architectures with considerable margins, which indicates that training small self-supervised contrastive models is feasible even without distillation signals. The code is available at \textit{https://github.com/WOWNICE/ssl-small}.