Beyond Loss Values: Robust Dynamic Pruning via Loss Trajectory Alignment
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866915924068532224 |
|---|---|
| author | Qin, Huaiyuan Yang, Muli Goenawan, Gabriel James Wang, Kai Wang, Zheng Hu, Peng Peng, Xi Zhu, Hongyuan |
| author_facet | Qin, Huaiyuan Yang, Muli Goenawan, Gabriel James Wang, Kai Wang, Zheng Hu, Peng Peng, Xi Zhu, Hongyuan |
| contents | Existing dynamic data pruning methods often fail under noisy-label settings, as they typically rely on per-sample loss as the ranking criterion. This could mistakenly lead to preserving noisy samples due to their high loss values, resulting in significant performance drop. To address this, we propose AlignPrune, a noise-robust module designed to enhance the reliability of dynamic pruning under label noise. Specifically, AlignPrune introduces the Dynamic Alignment Score (DAS), which is a loss-trajectory-based criterion that enables more accurate identification of noisy samples, thereby improving pruning effectiveness. As a simple yet effective plug-and-play module, AlignPrune can be seamlessly integrated into state-of-the-art dynamic pruning frameworks, consistently outperforming them without modifying either the model architecture or the training pipeline. Extensive experiments on five widely-used benchmarks across various noise types and pruning ratios demonstrate the effectiveness of AlignPrune, boosting accuracy by up to 6.3\% over state-of-the-art baselines. Our results offer a generalizable solution for pruning under noisy data, encouraging further exploration of learning in real-world scenarios. Code is available at: https://github.com/leonqin430/AlignPrune. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_07306 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Beyond Loss Values: Robust Dynamic Pruning via Loss Trajectory Alignment Qin, Huaiyuan Yang, Muli Goenawan, Gabriel James Wang, Kai Wang, Zheng Hu, Peng Peng, Xi Zhu, Hongyuan Computer Vision and Pattern Recognition Machine Learning Existing dynamic data pruning methods often fail under noisy-label settings, as they typically rely on per-sample loss as the ranking criterion. This could mistakenly lead to preserving noisy samples due to their high loss values, resulting in significant performance drop. To address this, we propose AlignPrune, a noise-robust module designed to enhance the reliability of dynamic pruning under label noise. Specifically, AlignPrune introduces the Dynamic Alignment Score (DAS), which is a loss-trajectory-based criterion that enables more accurate identification of noisy samples, thereby improving pruning effectiveness. As a simple yet effective plug-and-play module, AlignPrune can be seamlessly integrated into state-of-the-art dynamic pruning frameworks, consistently outperforming them without modifying either the model architecture or the training pipeline. Extensive experiments on five widely-used benchmarks across various noise types and pruning ratios demonstrate the effectiveness of AlignPrune, boosting accuracy by up to 6.3\% over state-of-the-art baselines. Our results offer a generalizable solution for pruning under noisy data, encouraging further exploration of learning in real-world scenarios. Code is available at: https://github.com/leonqin430/AlignPrune. |
| title | Beyond Loss Values: Robust Dynamic Pruning via Loss Trajectory Alignment |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2604.07306 |