本文介绍我们被AAAI'22接收的工作《On the Efficacy of Small Self-Supervised Contrastive Models without Distillation Signals》。该工作在OPPO研究院@Coler老师的指导下合作完成,同时向实习期间给予过宝贵帮助的同事们@油炸蘑菇表示感谢,尤其特别感谢前辈Zhiyuan Fang在课题初期给我们的建议以及code base的分享。
[1] Koohpayegani, S. A.; Tejankar, A.; and Pirsiavash, H. 2020. Compress: Self-supervised learning by compressing representations. arXiv preprint arXiv:2010.14713.
[2] Fang, Z.; Wang, J.; Wang, L.; Zhang, L.; Yang, Y.; and Liu, Z. 2021. Seed: Self-supervised distillation for visual representation. arXiv preprint arXiv:2101.04731.
[3] Xu, H.; Fang, J.; Zhang, X.; Xie, L.; Wang, X.; Dai, W.; Xiong, H.; and Tian, Q. 2021. Bag of Instances Aggregation Boosts Self-supervised Learning. arXiv preprint arXiv:2107.01691.
[4] u, J.; Liu, W.; and Tian, Y. 2021. Simple Distillation Baselines for Improving Small Self-supervised Models. arXiv preprint arXiv:2106.11304.
[5] Wang, F.; and Liu, H. 2021. Understanding the behaviour of contrastive loss. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2495–2504.
[6] Gao, Y.; Zhuang, J.-X.; Li, K.; Cheng, H.; Guo, X.; Huang, F.; Ji, R.; and Sun, X. 2021. DisCo: Remedy Self-supervised Learning on Lightweight Models with Distilled Contrastive Learning. arXiv preprint arXiv:2104.09124.
[7] Gao, T.; Yao, X.; and Chen, D. 2021. SimCSE: Simple Contrastive Learning of Sentence Embeddings. preprint arXiv:2104.08821.