aiqbench analysis
This paper tackles partial multi-label learning, where training data carries both correct and incorrect labels, a common result of crowd-sourced annotation. It proposes PML-CD, which derives an optimal curriculum weighting tied to model confidence calibration and learns that weighting with a transferable calibrator. Its practical value depends on whether the gains hold beyond the four benchmarks (Pascal VOC, MS-COCO, NUS-WIDE, CUB) and on how costly the calibrator is to train.
Calibrated Disambiguation for Partial Multi-label Learning
arXiv:2610.17635v1 Announce Type: new Abstract: Partial multi-label learning (PML) aims to train a classifier on datasets whose instances are over-annotated with not only relevant labels but also irrelevant labels, which is common when datasets are collected from crowd-sourcing platform. Existing works primarily approach it from a curriculum learning perspective, leveraging the memorization effect to disambiguate noisy labels and produce robust predictions. However, these methods are based on non-adaptive weighting functions and lack theoretical guidance for optimal weighting. To overcome these issues, a calibrated disambiguation model named PML-CD is proposed. We first formulate the optimal weighting function for curriculum-based disambiguation, which is equivalent to the calibration of the model's predicted confidences, thus providing guidance for curriculum designing. To obtain the optimal weighting function from PML dataset during the training, a transferable calibrator is designed, which takes the histogram of positive samples' confidences as input, and outputs the optimal curriculum weighting for training. Prototype alignment regularization is also proposed to promote the model's performance. Experiments conducted on Pascal VOC, MS-COCO, NUS-WIDE and CUB have verified that our method outperforms existing state-of-the-art PML methods.
Abstract:Partial multi-label learning (PML) aims to train a classifier on datasets whose instances are over-annotated with not only relevant labels but also irrelevant labels, which is common when datasets are collected from crowd-sourcing platform. Existing works primarily approach it from a curriculum learning perspective, leveraging the memorization effect to disambiguate noisy labels and produce robust predictions. However, these methods are based on non-adaptive weighting functions and lack theoretical guidance for optimal weighting. To overcome these issues, a calibrated disambiguation model named PML-CD is proposed. We first formulate the optimal weighting function for curriculum-based disambiguation, which is equivalent to the calibration of the model's predicted confidences, thus providing guidance for curriculum designing. To obtain the optimal weighting function from PML dataset during the training, a transferable calibrator is designed, which takes the histogram of positive samples' confidences as input, and outputs the optimal curriculum weighting for training. Prototype alignment regularization is also proposed to promote the model's performance. Experiments conducted on Pascal VOC, MS-COCO, NUS-WIDE and CUB have verified that our method outperforms existing state-of-the-art PML methods.
Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
arXiv:2610.17635 [cs.LG]
(or arXiv:2610.17635v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.17635 arXiv-issued DOI via DataCite (pending registration)
Zhuoming Li, Yuheng Jia, Mi Yu, Zicong Miao, "Calibrated disambiguation for partial multi-label learning." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 39. No. 17. 2025
https://doi.org/10.1609/aaai.v39i17.34049 DOI(s) linking to related resources
Full text under the publisher's open licence · 233 words
Related coverage
[AINews] not much happened today
Latent Space · Oct 3, 2026
CardioMeta: Calibrated Multi-Task Prediction of Diabetes, Hypertension, and Cardiovascular Disease Across Population and EHR Data
arXiv · Jul 20, 2026
CardioMeta: Calibrated Multi-Task Prediction of Diabetes, Hypertension, and Cardiovascular Disease Across Population and EHR Data
arXiv · Aug 25, 2026
Joint Class-Time Learning for Video Classification with Multi-Instance Partial-Label Learning
arXiv · Oct 7, 2026