arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2012.13670cs.LG
收藏

带噪声标签的鲁棒协作学习

Robust Collaborative Learning with Noisy Labels

Michigan State University(密歇根州立大学)

Mengying Sun, Jing Xing, Bin Chen, Jiayu Zhou

摘要:

在数据包含噪声(损坏)标签的任务中,使用课程进行学习已经显示出很好的效果,因为课程可以通过适当的设计来重新加权或过滤噪声样本。然而,从学习者本身获得课程,没有额外的监督或反馈恶化的有效性,由于样本选择的偏见。因此,最近提出了涉及两个或更多个网络的方法来减轻这种偏差。然而,这些研究利用网络之间的合作的方式,要么强调的分歧,或专注于协议,而忽略了其他。在本文中,我们研究了网络之间的分歧和协议如何帮助减少梯度中的噪声的潜在机制,并开发了一种称为鲁棒协作学习(RCL)的新框架,该框架利用了网络之间的分歧和协议。我们证明了RCL在合成基准图像数据和真实世界的大规模生物信息学数据上的有效性。

英文摘要:

Learning with curriculum has shown great effectiveness in tasks where the data contains noisy (corrupted) labels, since the curriculum can be used to re-weight or filter out noisy samples via proper design. However, obtaining curriculum from a learner itself without additional supervision or feedback deteriorates the effectiveness due to sample selection bias. Therefore, methods that involve two or more networks have been recently proposed to mitigate such bias. Nevertheless, these studies utilize the collaboration between networks in a way that either emphasizes the disagreement or focuses on the agreement while ignores the other. In this paper, we study the underlying mechanism of how disagreement and agreement between networks can help reduce the noise in gradients and develop a novel framework called Robust Collaborative Learning (RCL) that leverages both disagreement and agreement among networks. We demonstrate the effectiveness of RCL on both synthetic benchmark image data and real-world large-scale bioinformatics data.

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。

↑