aiqbench analysis
This research introduces a novel method for optimizing the partial AUC metric in binary classification tasks using only positive and unlabeled data, addressing the common scarcity of labeled negative samples due to privacy constraints. By deriving an empirical estimator within an empirical risk minimization framework, the approach enables effective model training for critical applications like cybersecurity and medical diagnostics without requiring explicit negative labels. The study validates its efficacy across ten real-world datasets, offering a practical solution for scenarios where negative data annotation is infeasible.
Partial AUC Maximization from Positive-unlabeled Data
arXiv:2610.00284v1 Announce Type: cross Abstract: The partial area under the receiver operating characteristic curve (pAUC) is an important performance metric for binary classification that summarizes true positive rates within a specific range of false positive rates (FPRs). Classifiers that achieve high pAUC need to be obtained in many real-world applications such as cybersecurity, medical care, and advertising. Although many methods for maximizing the pAUC have been proposed, they typically require both labeled positive and negative data for training. However, in practice, labeled negative data are often difficult to collect due to privacy concerns or the need for high expertise to annotate them. In this paper, we propose a method for maximizing the pAUC from positive and unlabeled (PU) data without negative data. Within an empirical risk minimization framework, we show that the pAUC, including its FPR-dependent thresholds, can be represented using only the positive and marginal densities, and derive an empirical estimator from PU data. A classifier is then trained by maximizing the derived smoothed empirical pAUC estimator. We experimentally demonstrate the effectiveness of the proposed method with ten real-world datasets.
Abstract:The partial area under the receiver operating characteristic curve (pAUC) is an important performance metric for binary classification that summarizes true positive rates within a specific range of false positive rates (FPRs). Classifiers that achieve high pAUC need to be obtained in many real-world applications such as cybersecurity, medical care, and advertising. Although many methods for maximizing the pAUC have been proposed, they typically require both labeled positive and negative data for training. However, in practice, labeled negative data are often difficult to collect due to privacy concerns or the need for high expertise to annotate them. In this paper, we propose a method for maximizing the pAUC from positive and unlabeled (PU) data without negative data. Within an empirical risk minimization framework, we show that the pAUC, including its FPR-dependent thresholds, can be represented using only the positive and marginal densities, and derive an empirical estimator from PU data. A classifier is then trained by maximizing the derived smoothed empirical pAUC estimator. We experimentally demonstrate the effectiveness of the proposed method with ten real-world datasets.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
arXiv:2610.00284 [cs.LG]
(or arXiv:2610.00284v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.00284 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 276 words
Related coverage
Deep Positive-Unlabeled Anomaly Detection for Contaminated Unlabeled Data
arXiv · Sep 26, 2026
AUC Maximization from Biased Positive-unlabeled Data with Confidence
arXiv · Sep 12, 2026
The art and science of hyperparameter optimization on Amazon Nova Forge
AWS · Jun 2, 2026
Minimal Vs Maximal superintelligence
LessWrong · Sep 23, 2026