L3A: Label-Augmented Analytic Adaptation for Multi-Label Class Incremental Learning

Xiang Zhang, Run He, Chen Jiao, Di Fang, Ming Li, Ziqian Zeng, Cen Chen, Huiping Zhuang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:74938-74949, 2025.

Abstract

Class-incremental learning (CIL) enables models to learn new classes continually without forgetting previously acquired knowledge. Multi-label CIL (MLCIL) extends CIL to a real-world scenario where each sample may belong to multiple classes, introducing several challenges: label absence, which leads to incomplete historical information due to missing labels, and class imbalance, which results in the model bias toward majority classes. To address these challenges, we propose Label-Augmented Analytic Adaptation (L3A), an exemplar-free approach without storing past samples. L3A integrates two key modules. The pseudo-label (PL) module implements label augmentation by generating pseudo-labels for current phase samples, addressing the label absence problem. The weighted analytic classifier (WAC) derives a closed-form solution for neural networks. It introduces sample-specific weights to adaptively balance the class contribution and mitigate class imbalance. Experiments on MS-COCO and PASCAL VOC datasets demonstrate that L3A outperforms existing methods in MLCIL tasks. Our code is available at https://github.com/scut-zx/L3A.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-zhang25y, title = {{L}3{A}: Label-Augmented Analytic Adaptation for Multi-Label Class Incremental Learning}, author = {Zhang, Xiang and He, Run and Jiao, Chen and Fang, Di and Li, Ming and Zeng, Ziqian and Chen, Cen and Zhuang, Huiping}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {74938--74949}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/zhang25y/zhang25y.pdf}, url = {https://proceedings.mlr.press/v267/zhang25y.html}, abstract = {Class-incremental learning (CIL) enables models to learn new classes continually without forgetting previously acquired knowledge. Multi-label CIL (MLCIL) extends CIL to a real-world scenario where each sample may belong to multiple classes, introducing several challenges: label absence, which leads to incomplete historical information due to missing labels, and class imbalance, which results in the model bias toward majority classes. To address these challenges, we propose Label-Augmented Analytic Adaptation (L3A), an exemplar-free approach without storing past samples. L3A integrates two key modules. The pseudo-label (PL) module implements label augmentation by generating pseudo-labels for current phase samples, addressing the label absence problem. The weighted analytic classifier (WAC) derives a closed-form solution for neural networks. It introduces sample-specific weights to adaptively balance the class contribution and mitigate class imbalance. Experiments on MS-COCO and PASCAL VOC datasets demonstrate that L3A outperforms existing methods in MLCIL tasks. Our code is available at https://github.com/scut-zx/L3A.} }
Endnote
%0 Conference Paper %T L3A: Label-Augmented Analytic Adaptation for Multi-Label Class Incremental Learning %A Xiang Zhang %A Run He %A Chen Jiao %A Di Fang %A Ming Li %A Ziqian Zeng %A Cen Chen %A Huiping Zhuang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-zhang25y %I PMLR %P 74938--74949 %U https://proceedings.mlr.press/v267/zhang25y.html %V 267 %X Class-incremental learning (CIL) enables models to learn new classes continually without forgetting previously acquired knowledge. Multi-label CIL (MLCIL) extends CIL to a real-world scenario where each sample may belong to multiple classes, introducing several challenges: label absence, which leads to incomplete historical information due to missing labels, and class imbalance, which results in the model bias toward majority classes. To address these challenges, we propose Label-Augmented Analytic Adaptation (L3A), an exemplar-free approach without storing past samples. L3A integrates two key modules. The pseudo-label (PL) module implements label augmentation by generating pseudo-labels for current phase samples, addressing the label absence problem. The weighted analytic classifier (WAC) derives a closed-form solution for neural networks. It introduces sample-specific weights to adaptively balance the class contribution and mitigate class imbalance. Experiments on MS-COCO and PASCAL VOC datasets demonstrate that L3A outperforms existing methods in MLCIL tasks. Our code is available at https://github.com/scut-zx/L3A.
APA
Zhang, X., He, R., Jiao, C., Fang, D., Li, M., Zeng, Z., Chen, C. & Zhuang, H.. (2025). L3A: Label-Augmented Analytic Adaptation for Multi-Label Class Incremental Learning. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:74938-74949 Available from https://proceedings.mlr.press/v267/zhang25y.html.

Related Material