通过特征相似性自动识别并修正数据标签错误，提升模型训练质量。
Detecting and Rectifying Noisy Labels: A Similarity-based Approach
- 利用深层网络倒数第二层特征的相似性判断标签是否错误。
- 在多种真实噪声场景下检测准确率高，可自动修正错误标签。
- 无需修改模型，适用于各类深度神经网络，适合数据清洗场景。
数据集中的标签噪声会严重损害深度神经网络（DNN）的性能与鲁棒性。随着现代DNN规模不断扩大，对自动化标签错误检测工具的需求日益增长。本文提出一种后置、模型无关的噪声检测与修正方法，利用DNN的倒数第二层特征。核心思想是：错误标签样本的特征与其真实类别样本的特征相似度高于其他类别，因此在紧密相似簇内标签出现的概率可作为检测和修正错误的依据。通过理论与实证分析，我们证明该方法在多种真实噪声场景下具有优异的检测性能，并能自动修正错误以提升数据集质量。代码已公开于 https://anonymous.4open.science/r/noise-detection-and-rectification-AD8E。
原文摘要 · Abstract (English)
Label noise in datasets could significantly damage the performance and robustness of deep neural networks (DNNs) trained on these datasets. As the size of modern DNNs grows, there is a growing demand for automated tools for detecting such errors. In this paper, we propose post-hoc, model-agnostic noise detection and rectification methods utilizing the penultimate feature from a DNN. Our idea is based on the observation that the similarity between the penultimate feature of a mislabeled data point and its true class data points is higher than that for data points from other classes, making the probability of label occurrence within a tight, similar cluster informative for detecting and rectifying errors. Through theoretical and empirical analyses, we demonstrate that our approach achieves high detection performance across diverse, realistic noise scenarios and can automatically rectify these errors to improve dataset quality. Our implementation is available at https://anonymous.4open.science/r/noise-detection-and-rectification-AD8E.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。