[1] Babakhin, Y., Sanakoyeu, A., & Kitamura, H. (2019). Semi-supervised segmentation of salt bodies in seismic images using an ensemble of convolutional neural networks. ArXiv:1904.04445. http://arxiv.org/abs/1904.04445
- Paper describing the winning solution to the Kaggle TGS Salt Identification Challenge including pseudolabelling
- Also see Yauhen Babakhin's at Kaggle Days titled How to cook pseudo-labels
[2] Internet.org and Facebook (2013). A Focus on Efficiency. [White paper]. Archived on webarchive.org
[3] Li, Z., Ko, B., & Choi, H.-J. (2019). Naive semi-supervised deep learning using pseudo-label. Peer-to-Peer Networking and Applications, 12(5), 1358–1368. https://doi.org/10.1007/s12083-018-0702-9
- Paper describing pretrain method for pseudolabelling as well as results from LSTM, CIFAR, MNIST.
[4] Xie, Q., Luong, M.-T., Hovy, E., & Le, Q. V. (2020). Self-training with Noisy Student improves ImageNet classification. ArXiv:1911.04252 [Cs, Stat]. http://arxiv.org/abs/1911.04252
[5] Global Wheat competition: https://www.kaggle.com/c/global-wheat-detection
- GitHub repo with all code for performance presented in the talk: https://github.com/stanleyjzheng/Global-Wheat
[6] The following are solutions from the mentioned competitions: OpenVaccine 1st, OpenVaccine 2nd, OpenVaccine 3rd, Tweet sentiment extraction 1st, TReNDS Neuroimaging 1st, Global Wheat 1st, LISH-MOA 2nd public, LISH-MOA 5th, TGS Salt Identification 1st
[7] My notebook on pseudolabelling MNIST https://www.kaggle.com/stanleyjzheng/exploring-pseudolabelling-schemes-pydata