arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.07756cs.LGcs.CVstat.APstat.ML
收藏

使用依赖于时间的标签噪声学习:一种渐进方法

Learning with Feature-Dependent Label Noise: A Progressive Approach

Rutgers University(罗格斯大学);Stony Brook University(石溪大学);City University of New York(纽约城市大学)

Yikai Zhang, Songzhu Zheng, Pengxiang Wu, Mayank Goswami, Chao Chen

摘要:

标签噪声在现实世界的大规模数据集中经常被观察到。噪声是由于各种原因而引入的;它是异质的并且依赖于特征。大多数现有的方法来处理嘈杂的标签分为两类:他们要么假设一个理想的功能独立的噪音,或保持启发式没有理论保证。在本文中,我们提出了一个新的家庭的特征相关的标签噪声,这是更普遍的比常用的i.i.d.。标记噪声并包含广泛的噪声模式。针对这个一般的噪声族,我们提出了一个渐进的标签校正算法,迭代地校正标签和细化模型。我们提供的理论保证表明,对于各种各样的(未知的)噪声模式,用这种策略训练的分类器收敛到与贝叶斯分类器一致。在实验中,我们的方法优于SOTA基线,并且对各种噪声类型和水平具有鲁棒性。

英文摘要:

Label noise is frequently observed in real-world large-scale datasets. The noise is introduced due to a variety of reasons; it is heterogeneous and feature-dependent. Most existing approaches to handling noisy labels fall into two categories: they either assume an ideal feature-independent noise, or remain heuristic without theoretical guarantees. In this paper, we propose to target a new family of feature-dependent label noise, which is much more general than commonly used i.i.d. label noise and encompasses a broad spectrum of noise patterns. Focusing on this general noise family, we propose a progressive label correction algorithm that iteratively corrects labels and refines the model. We provide theoretical guarantees showing that for a wide variety of (unknown) noise patterns, a classifier trained with this strategy converges to be consistent with the Bayes classifier. In experiments, our method outperforms SOTA baselines and is robust to various noise types and levels.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑