Boosting Learning Efficiency in Few-Shot Tasks With Layer-Adaptive PID Control
IEEE Transactions on Pattern Analysis and Machine Intelligence · 2026-06 · 48 (6) · 7079-7096
DOI: 10.1109/tpami.2026.3663608 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026-06. Verfasst von Pengfei Zhang und 6 weiteren Autor:innen. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 99 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Key Projects of Key R&D Program of Jiangsu Province, Shenzhen Science and Technology Program.
Abstract (Originalsprache)
Few-shot learning seeks to recognize novel classes from limited examples. Model-agnostic meta-learning (MAML), known for its simplicity and flexibility, learns an effective initialization for fast adaptation in data-scarce settings. However, MAML-based methods face challenges when there is a significant distributional shift between training and testing tasks, leading to inefficient learning and poor generalization across domains. In this work, we identify the core issues: inflexible weight update rules and limited adaptive learning capabilities. Instead of focusing solely on better initialization, we aim to enhance the adaptation process. Consequently, we propose a novel Layer-Adaptive Proportional-Integral-Derivative (LA-PID) optimizer integrated into a meta-learning framework. This design incorporates classical control theory, utilizing PID control to dynamically adjust task-specific gains at each network layer. Additionally, the theoretical conditions for optimal hyperparameter initialization and global model convergence are addressed from both control and optimization perspectives. Experiments on benchmark datasets show that LA-PID achieves state-of-the-art performance in few-shot classification, cross-domain, and regression tasks, while requiring fewer training steps.
Förderung
- National Natural Science Foundation of China
- Key Projects of Key R&D Program of Jiangsu Province
- Shenzhen Science and Technology Program
Zitieren
APA
Zhang, P., Li, X., Yu, L., Zhang, Z., Dunkin, F., Liu, H., & Li, Z. (2026). Boosting Learning Efficiency in Few-Shot Tasks With Layer-Adaptive PID Control. IEEE Transactions on Pattern Analysis and Machine Intelligence, 48(6), 7079-7096. https://doi.org/10.1109/tpami.2026.3663608
Vancouver
Zhang P, Li X, Yu L, Zhang Z, Dunkin F, Liu H, et al. Boosting Learning Efficiency in Few-Shot Tasks With Layer-Adaptive PID Control. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2026;48(6):7079-7096. doi:10.1109/tpami.2026.3663608
BibTeX
@article{Zhang2026,
author = {Zhang, Pengfei and Li, Xinde and Yu, Le and Zhang, Zhentong and Dunkin, Fir and Liu, Huaping and Li, Zhijun},
title = {Boosting Learning Efficiency in Few-Shot Tasks With Layer-Adaptive PID Control},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
year = {2026},
volume = {48},
number = {6},
pages = {7079--7096},
doi = {10.1109/tpami.2026.3663608}
}
EUVIMED