arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.08736cs.LGcs.AIcs.CVmath.OC
收藏

具有可证明收敛性的精确率-召回率曲线下面积随机优化

Stochastic Optimization of Areas Under Precision-Recall Curves with Provable Convergence

The University of Iowa(爱荷华大学);Texas A&M University(得克萨斯农工大学)

Qi Qi, Youzhi Luo, Zhao Xu, Shuiwang Ji, Tianbao Yang

AI总结:

本文针对高度不平衡数据集的AUPRC优化问题,提出基于平均精确率最大化的随机算法SOAP,首次实现AUPRC的可证明收敛优化,并在图像和图数据集上验证了其优越性。

摘要:

ROC曲线下面积(AUROC)和精确率-召回率曲线下面积(AUPRC)是评估不平衡问题分类性能的常用指标。与AUROC相比,AUPRC对于高度不平衡的数据集是更合适的指标。虽然AUROC的随机优化已被广泛研究,但有原则的AUPRC随机优化却很少被探索。在这项工作中,我们提出了一种有原则的技术方法来优化深度学习中的AUPRC。我们的方法基于最大化平均精确率(AP),它是AUPRC的无偏点估计量。我们将目标转化为若干依赖的组合函数之和,其中内层函数依赖于外层随机变量。通过利用随机组合优化的最新进展,我们提出了名为SOAP的高效自适应和非自适应随机算法,在温和条件下具有可证明的收敛保证。在图像和图数据集上的大量实验结果表明,我们提出的方法在不平衡问题上以AUPRC为指标优于先前的方法。据我们所知,我们的工作代表了首次尝试以可证明的收敛性优化AUPRC。SOAP已在libAUC库中实现,网址为https://libauc.org/。

英文摘要:

Areas under ROC (AUROC) and precision-recall curves (AUPRC) are common metrics for evaluating classification performance for imbalanced problems. Compared with AUROC, AUPRC is a more appropriate metric for highly imbalanced datasets. While stochastic optimization of AUROC has been studied extensively, principled stochastic optimization of AUPRC has been rarely explored. In this work, we propose a principled technical method to optimize AUPRC for deep learning. Our approach is based on maximizing the averaged precision (AP), which is an unbiased point estimator of AUPRC. We cast the objective into a sum of {\it dependent compositional functions} with inner functions dependent on random variables of the outer level. We propose efficient adaptive and non-adaptive stochastic algorithms named SOAP with {\it provable convergence guarantee under mild conditions} by leveraging recent advances in stochastic compositional optimization. Extensive experimental results on image and graph datasets demonstrate that our proposed method outperforms prior methods on imbalanced problems in terms of AUPRC. To the best of our knowledge, our work represents the first attempt to optimize AUPRC with provable convergence. The SOAP has been implemented in the libAUC library at~\url{https://libauc.org/}.

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。

↑