Sharpness-Aware Dynamic Anchor Selection for Generalized Category Discovery
- 15 January 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Multimedia
- Vol. 28 (15209210) , 3613-3623
Abstract
Generalized category discovery (GCD) is an important and challenging task in open-world learning. Specifically, given some labeled data of known classes, GCD aims to cluster unlabeled data that contain both known and unknown classes. Current GCD methods based on parametric classification adopt the DINO-like pseudo-labeling strategy, where the sharpened probability output of one view is used as supervision information for the other view. However, large pre-trained models have a preference for some specific visual patterns, resulting in encoding spurious correlation for unlabeled data and generating noisy pseudo-labels. To address this issue, we propose a novel method, which contains two modules: Loss Sharpness Penalty (LSP) and Dynamic Anchor Selection (DAS). LSP enhances the robustness of model parameters to small perturbations by minimizing the worst-case loss sharpness of the model, which suppressing the encoding of trivial features, thereby reducing overfitting of noise samples and improving the quality of pseudo-labels. Meanwhile, DAS selects representative samples for the unknown classes based on KNN density and class probability during the model training and assigns hard pseudo-labels to them, which not only alleviates the confidence difference between known and unknown classes but also enables the model to quickly learn more accurate feature distribution for the unknown classes, thus further improving the clustering accuracy. Extensive experiments demonstrate that the proposed method can effectively mitigate the noise of pseudo-labels, and achieve state-of-the-art results on multiple GCD benchmarks.Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62225604)
- National Natural Science Foundation of China (62176130)
- Tianjin Natural Science (24JCQNJC01850)
This publication has 35 references indexed in Scilit:
- SelEx: Self-expertise in Fine-Grained Generalized Category Discovery2024
- Unsupervised Selective Labeling for More Effective Semi-supervised Learning2022
- Rethinking Confidence Calibration for Failure Prediction2022
- Towards Efficient and Scalable Sharpness-Aware Minimization2022
- Generalized Category Discovery2022
- Neighborhood Contrastive Learning for Novel Class Discovery2021
- Regularizing Neural Networks via Adversarial Model Perturbation2021
- OpenMix: Reviving Known Knowledge for Discovering Novel Visual Categories in an Open World2021
- Deep Residual Learning for Image Recognition2016
- The Hungarian method for the assignment problemNaval Research Logistics Quarterly, 1955