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 2302.12987cs.LG
收藏

对多标签的互补性:一种感知相关性的校正方法

Complementary to Multiple Labels: A Correlation-Aware Correction Approach

Southeast University(东南大学);The University of Queensland(昆士兰大学)

Yi Gao, Miao Xu, Min-Ling Zhang

AI总结:

针对多类互补标签学习无法适配多标签数据的问题,提出基于二分类分解加标签相关性校正的两步转移矩阵估计法,证明其分类器一致性并引入MSE正则项缓解过拟合,实验验证了方法有效性。

摘要:

互补标签学习(Complementary label learning, CLL)要求标注者为样本给出不相关标签而非相关标签。目前,CLL通过估计转移矩阵已在多类数据上展现出良好性能。然而,现有的多类CLL技术无法很好地应用于多标签数据,因为它们假设每个样本仅关联一个标签,而每个多标签样本都与多个标签相关。本文从理论上证明,由于多类CLL忽略了共存的相关标签,其估计的转移矩阵在多标签场景下会出现失真。此外,理论发现表明,在多标签互补标签学习(multi-labeled CLL, ML-CLL)中,根据标签相关性计算转移矩阵需要用到多标签数据,但这在ML-CLL场景下是无法获取的。为解决该问题,本文提出一种两步法从候选标签估计转移矩阵。具体而言,首先通过将多标签问题分解为一系列二分类问题来估计初始转移矩阵,随后利用标签相关性校正初始转移矩阵,以引入标签间的关联关系。本文进一步证明了所提方法具有分类器一致性,还引入了一种基于均方误差(MSE)的正则项,以缓解二元交叉熵(BCE)损失对噪声过拟合的趋势。实验结果证明了所提方法的有效性。

英文摘要:

\textit{Complementary label learning} (CLL) requires annotators to give \emph{irrelevant} labels instead of relevant labels for instances. Currently, CLL has shown its promising performance on multi-class data by estimating a transition matrix. However, current multi-class CLL techniques cannot work well on multi-labeled data since they assume each instance is associated with one label while each multi-labeled instance is relevant to multiple labels. Here, we show theoretically how the estimated transition matrix in multi-class CLL could be distorted in multi-labeled cases as they ignore co-existing relevant labels. Moreover, theoretical findings reveal that calculating a transition matrix from label correlations in \textit{multi-labeled CLL} (ML-CLL) needs multi-labeled data, while this is unavailable for ML-CLL. To solve this issue, we propose a two-step method to estimate the transition matrix from candidate labels. Specifically, we first estimate an initial transition matrix by decomposing the multi-label problem into a series of binary classification problems, then the initial transition matrix is corrected by label correlations to enforce the addition of relationships among labels. We further show that the proposal is classifier-consistent, and additionally introduce an MSE-based regularizer to alleviate the tendency of BCE loss overfitting to noises. Experimental results have demonstrated the effectiveness of the proposed method.

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。

↑