We introduce a dynamic temperature scaling mechanism for self-supervised contrastive learning to improve representation quality across varied data distributions.
@article{manna2025dynamically,
title={Dynamically Scaled Temperature in Self-Supervised Contrastive Learning},
author={Manna, S and Chattopadhyay, S and Dey, R and Pal, U and Bhattacharya, S},
journal={IEEE Transactions on Artificial Intelligence},
year={2025},
doi={10.1109/TAI.2024.3456789},
url={https://sadimanna.github.io/papers/dyst-ssl/}
}