Dynamic LoRA-Experts and Prototype-Ensemble Matching for Class-Incremental Learning
Applied Sciences · 17.06.2026 · 16 (12) · 6153
DOI: 10.3390/app16126153 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Sciences, online erschienen am 17.06.2026. Verfasst von Hongwei Zhao und 2 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Class-incremental learning (CIL) aims to continuously learn new classes without forgetting previously acquired knowledge. Recent advances in parameter-efficient fine-tuning (PEFT) based on pre-trained models (PTMs) have shown promise in this setting by integrating new tasks with minimal parameter overhead. However, these methods often suffer from knowledge degradationdue to: (1) cumulative interference caused by iterative updates, constrained gradient flows, or entangled module integration; and (2) suboptimal alignment between inference samples and specialized modules. To address these challenges, we propose Dynamic LoRA-Experts and Prototype-Ensemble Matching (DLEPEM), a novel two-stage, rehearsal-free framework. In the first stage, we allocate a task-specific LoRA-Expert for each incremental task, enabling isolated representation learning and reducing cross-task interference. In the second stage, we introduce a prototype-ensemble-matching mechanism that combines general prototypes derived from the frozen PTM with task-adaptive prototypes learned by the LoRA-Experts. This fusion facilitates both strong generalization and precise task-level discrimination. Extensive experiments on standard CIL and few-shot class-incremental learning (FSCIL) benchmarks demonstrate that DLEPEM achieves strong performance under the evaluated protocols. For instance, in CIL, it achieves 93.39% on CIFAR100 (+0.80% over EASE), 92.31% on CUB200 (+2.11% over EASE), and 91.84% on VTAB (+1.39% over EASE). In the more challenging FSCIL setting, it achieves 88.77% on CUB200, outperforming the strongest baseline by a clear margin of 5.31%. These results indicate that DLEPEM effectively mitigates catastrophic forgetting while enhancing incremental learning capability.
Zitieren
APA
Zhao, H., Liu, R., & Liu, Y. (2026). Dynamic LoRA-Experts and Prototype-Ensemble Matching for Class-Incremental Learning. Applied Sciences, 16(12), 6153. https://doi.org/10.3390/app16126153
Vancouver
Zhao H, Liu R, Liu Y. Dynamic LoRA-Experts and Prototype-Ensemble Matching for Class-Incremental Learning. Applied Sciences. 2026;16(12):6153. doi:10.3390/app16126153
BibTeX
@article{Zhao2026,
author = {Zhao, Hongwei and Liu, Rui and Liu, Yansong},
title = {Dynamic LoRA-Experts and Prototype-Ensemble Matching for Class-Incremental Learning},
journal = {Applied Sciences},
year = {2026},
volume = {16},
number = {12},
pages = {6153},
doi = {10.3390/app16126153}
}
EUVIMED