VALSE 论文速览 第134期：基于解耦元学习标签纠正的噪声标签学习 ...

论文题目：Learning from Noisy Labels with Decoupled Meta Label Purifier

作者列表：

涂远鹏 (同济大学),张博深 (腾讯优图)，李昱希 (腾讯优图)，刘亮 (腾讯优图)，李剑 (腾讯优图)，王亚彪 (腾讯优图)，汪铖杰 (腾讯优图)，赵才荣 (同济大学)

B站观看网址：

https://www.bilibili.com/video/BV13w411A7Nk/

论文摘要：

Training deep neural networks (DNN) with noisy labels is challenging since DNN can easily memorize inaccurate labels, leading to poor generalization ability. Recently, the meta-learning based label correction strategy is widely adopted to tackle this problem via identifying and correcting potential noisy labels with the help of a small set of clean validation data. Although training with purified labels can effectively improve performance, solving the meta-learning problem inevitably involves a nested loop of bi-level optimization between model weights and hyper-parameters (i.e., label distribution). As compromise, previous methods resort to a coupled learning process with alternating update. In this paper, we empirically find such simultaneous optimization over both model weights and label distribution can not achieve an optimal routine, consequently limiting the representation ability of backbone and accuracy of corrected labels. From this observation, a novel multi-stage label purifier named DMLP is proposed. DMLP decouples the label correction process into label-free representation learning and a simple meta label purifier. In this way, DMLP can focus on extracting discriminative feature and label correction in two distinctive stages. DMLP is a plug-and-play label purifier, the purified labels can be directly reused in naive end-to-end network retraining or other robust learning methods, where state-of-the-art results are obtained on several synthetic and real-world noisy datasets, especially under high noise levels.

论文信息：

[1] Yuanpeng Tu , Boshen Zhang, Yuxi Li, Liang Liu, Jian Li, Yabiao Wang, Chengjie Wang, Cai Rong Zhao, “Learning from Noisy Labels with Decoupled Meta Label Purifier,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR2023), Vancouver Convention Center, Jun, 2023.

论文链接：

[https://arxiv.org/abs/2302.06810]

代码链接：

[https://github.com/yuanpengtu/DMLP]

视频讲者简介：

涂远鹏，同济大学电子与信息工程学院硕士生。主要研究为行人再识别与噪声标签学习。

个人主页：

https://yuanpengtu.github.io/

同济大学视觉与智能学习实验室主页：

https://vill.tongji.edu.cn/

特别鸣谢本次论文速览主要组织者：

月度轮值AC：叶茫 (武汉大学)