38 nlnl negative learning for noisy labels

伪标签还能这样用？半监督力作UPS（ICLR 21）大揭秘！ - 知乎此外，作者还顺着带噪学习的藤，摸到了Negative Learning的瓜：如图3所示，我们虽不知道样本属于哪类，但对它不属于哪类还是蛮有把握的（Negative learning for noisy labels，ICCV 2019）。这样的伪标签相比传统的Positive Learning的伪标签更为准确，因而能很好地降低标签的 ... GitHub - subeeshvasu/Awesome-Learning-with-Label-Noise: A ... 2019-ICCV - NLNL: Negative Learning for Noisy Labels. 2019-ICCV - Symmetric Cross Entropy for Robust Learning With Noisy Labels. 2019-ICCV - Co-Mining: Deep Face Recognition With Noisy Labels. 2019-ICCV - O2U-Net: A Simple Noisy Label Detection Approach for Deep Neural Networks.