TLCE: Transfer-Learning Based Classifier Ensembles for Few-Shot Class-Incremental Learning
Neural Processing Letters · 08.05.2024 · 56 (3)
DOI: 10.1007/s11063-024-11605-0 ↗Kurz gefasst
Fachartikel in der Zeitschrift Neural Processing Letters, online erschienen am 08.05.2024. Verfasst von Shuangmei Wang und 2 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Few-shot class-incremental learning (FSCIL) struggles to incrementally recognize novel classes from few examples without catastrophic forgetting of old classes or overfitting to new classes. We propose TLCE, which ensembles multiple pre-trained models to improve separation of novel and old classes. Specifically, we use episodic training to map images from old classes to quasi-orthogonal prototypes, which minimizes interference between old and new classes. Then, we incorporate the use of ensembling diverse pre-trained models to further tackle the challenge of data imbalance and enhance adaptation to novel classes. Extensive experiments on various datasets demonstrate that our transfer learning ensemble approach outperforms state-of-the-art FSCIL methods.
Zitieren
APA
Wang, S., Cao, Y., & Wu, T. (2024). TLCE: Transfer-Learning Based Classifier Ensembles for Few-Shot Class-Incremental Learning. Neural Processing Letters, 56(3). https://doi.org/10.1007/s11063-024-11605-0
Vancouver
Wang S, Cao Y, Wu T. TLCE: Transfer-Learning Based Classifier Ensembles for Few-Shot Class-Incremental Learning. Neural Processing Letters. 2024;56(3). doi:10.1007/s11063-024-11605-0
BibTeX
@article{Wang2024,
author = {Wang, Shuangmei and Cao, Yang and Wu, Tieru},
title = {TLCE: Transfer-Learning Based Classifier Ensembles for Few-Shot Class-Incremental Learning},
journal = {Neural Processing Letters},
year = {2024},
volume = {56},
number = {3},
doi = {10.1007/s11063-024-11605-0}
}
EUVIMED