Unsupervised label noise modeling and loss correction
Unsupervised label noise modeling and loss correction
复制标题
DOI：
--
复制
发表时间：
2019-04
期刊：
影响因子：
--
通讯作者：
Eric Arazo Sanchez;Diego Ortego;Paul Albert;N. O’Connor;Kevin McGuinness
中科院分区：
文献类型：
--
作者：
Eric Arazo Sanchez;Diego Ortego;Paul Albert;N. O’Connor;Kevin McGuinness
尽管对少量标签噪声具有鲁棒性，但使用随机梯度方法训练的卷积神经网络已被证明可以轻松拟合随机标签。当有正确和错误标记的目标混合时，网络倾向于在后者之前拟合前者。这表明在训练期间使用合适的双组分混合模型作为样本损失值的无监督生成模型，以允许在线估计样本被错误标记的概率。具体来说，我们提出了一个beta混合来估计这个概率，并通过依赖网络预测来纠正损失（所谓的自举损失）。我们进一步调整混合增强，以进一步推动我们的方法。在CIFAR-10/100和TinyImageNet上的实验证明了对标签噪声的鲁棒性，其性能大大优于最新的最先进技术。源代码可在https://git.io/fjsvE和附录https://arxiv.org/abs/1904.11238上获得。
Despite being robust to small amounts of label noise, convolutional neural networks trained with stochastic gradient methods have been shown to easily fit random labels. When there are a mixture of correct and mislabelled targets, networks tend to fit the former before the latter. This suggests using a suitable two-component mixture model as an unsupervised generative model of sample loss values during training to allow online estimation of the probability that a sample is mislabelled. Specifically, we propose a beta mixture to estimate this probability and correct the loss by relying on the network prediction (the so-called bootstrapping loss). We further adapt mixup augmentation to drive our approach a step further. Experiments on CIFAR-10/100 and TinyImageNet demonstrate a robustness to label noise that substantially outperforms recent state-of-the-art. Source code is available at https://git.io/fjsvE and Appendix at https://arxiv.org/abs/1904.11238.