Detecting Corrupted Labels Without Training a Model to Predict
Detecting Corrupted Labels Without Training a Model to Predict
复制标题
DOI：
--
复制
发表时间：
2021-10
期刊：
影响因子：
--
通讯作者：
Zhaowei Zhu;Zihao Dong;Yang Liu
中科院分区：
文献类型：
--
作者：
Zhaowei Zhu;Zihao Dong;Yang Liu
现实数据集中的标签噪声编码错误的相关模式，损害了深度神经网络（dnn）的泛化。找到检测损坏模式的有效方法至关重要。目前的方法主要集中在设计健壮的训练技术来防止dnn记忆损坏的模式。这些方法通常需要定制的训练过程，并且可能会过度拟合损坏的模式，从而导致检测性能下降。在本文中，从更以数据为中心的角度，我们提出了一种无需训练的解决方案来检测损坏的标签。直觉上，“更接近”的实例更有可能共享相同的干净标签。基于邻域信息，我们提出了两种方法：第一种方法是通过检查邻域特征的噪声标签共识来进行“局部投票”。第二种方法是基于排名的方法，对每个实例进行评分，并过滤出可能损坏的实例的保证数量。我们从理论上分析了特征的质量如何影响局部投票，并提供了调整邻域大小的指导方针。我们还证明了基于排序方法的最坏情况误差界。合成和真实世界标签噪声的实验表明，我们的无训练解决方案一致且显著地改善了大多数基于训练的基线。代码可从github.com/UCSC-REAL/SimiFeat获得。
Label noise in real-world datasets encodes wrong correlation patterns and impairs the generalization of deep neural networks (DNNs). It is critical to find efficient ways to detect corrupted patterns. Current methods primarily focus on designing robust training techniques to prevent DNNs from memorizing corrupted patterns. These approaches often require customized training processes and may overfit corrupted patterns, leading to a performance drop in detection. In this paper, from a more data-centric perspective, we propose a training-free solution to detect corrupted labels. Intuitively, ``closer'' instances are more likely to share the same clean label. Based on the neighborhood information, we propose two methods: the first one uses ``local voting"via checking the noisy label consensuses of nearby features. The second one is a ranking-based approach that scores each instance and filters out a guaranteed number of instances that are likely to be corrupted. We theoretically analyze how the quality of features affects the local voting and provide guidelines for tuning neighborhood size. We also prove the worst-case error bound for the ranking-based method. Experiments with both synthetic and real-world label noise demonstrate our training-free solutions consistently and significantly improve most of the training-based baselines. Code is available at github.com/UCSC-REAL/SimiFeat.