1.4 《Learning with Bounded Instance- and label-dependent Label Noise》 This paper focus on Bounded Instance- and Label-dependent label Noise (BILN), a particular case of Label-dependent label Noise where the label noise rates. This paper focus on a particular case of ILN where noise rates h...
CVPR2019: Probabilistic End-to-end Noise Correction for Learning with Noisy Labels 对上篇文章的改进 用模型去学习样本真实标签的分布,Joint Training,不再做两步更新 1.2 Dataset Pruning 直接移除噪声数据,同样可以达到“清洗数据,使用干净数据训练分类器”的目的 ICCV2019: O2U-Net: A Simple Noisy Label Dete...
A curated list of resources for Learning with Noisy Labels - congyang1996/Awesome-Learning-with-Label-Noise
A curated list of resources for Learning with Noisy Labels - JUNXYU/Awesome-Learning-with-Label-Noise
Learning with symmetric label noise: The importance of being unhinged. In NIPS*28, 2015.B. van Rooyen, A. K. Menon, and R. C. Williamson. Learning with symmetric label noise: The importance of being unhinged. In NIPS*29, 2015.
The correct unknown label can be viewed as a hidden random variable Model the noise processes by a communication channel with unknown parameters. 用EM 算法找 network 和 correct label。这个思路 2012,2016 年都有不错的文章发...
However, Long and Servedio [2010] proved that under symmetric label noise (SLN), minimisation of any convex potential over a linear function class can result in classification performance equivalent to random guessing. This ostensibly shows that convex losses are not SLN-robust. In this paper, we...
损失调整:包括损失校正、损失重加权、标签翻新(label refurbishment)和元学习 样本选择:【有点类似数据清洗的想法?】 A. Robust Architecture (1)噪声自适应层(noise adaptation layer) 专门用噪声过渡矩阵来建模噪声转换,测试的去掉这一层。只针对实例无关的噪声。
However, we observe that the advantage of LS vanishes when we operate in a high label noise regime. Puzzled by the observation, we proceeded to discover that several proposed learning-with-noisy-labels solutions in the literature instead relate more closely to negative label smoothing (NLS), ...
(一) Learning with label noise 有一种方法使用DNN、概率图模型、知识图谱或条件随机场等方式建立显式或隐式噪声模型来表征噪声和真实标签的分布;对噪声建模的结果用来把修正标签或者给噪声标签赋予更小的权重。但是这种方法通常需要小部分干净的标签子集,或者要用比较expensive的评估准则。除此之外,这种方法不能应用于...