Research on fine-tuning algorithms for Large Language Models integrating Uncertainty Modeling and External Memory Augmentation
PLOS One · 2026
DOI: 10.1371/journal.pone.0351493 ↗Kurz gefasst
Fachartikel in der Zeitschrift PLOS One, 2026. Verfasst von Yumeng Ma und 7 weiteren Autor:innen. Beteiligte Einrichtung: Arizona State University, Tempe, Arizona, United States of America. Hauptthemen laut MEDLINE: Algorithms, Natural Language Processing, Memory. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This paper proposes a parameter-efficient fine-tuning framework that integrates uncertainty modeling with external memory augmentation, aiming to improve robustness, confidence calibration, and contextual completeness in downstream natural language processing tasks. From the methodological perspective, the uncertainty modeling module explicitly characterizes uncertainty in inputs and intermediate representations through feature-level estimation, cross-layer propagation, and confidence calibration, thereby enhancing training stability and reducing the influence of noisy signals. Meanwhile, the external memory augmentation module employs key-value retrieval and gated fusion mechanisms to provide reusable contextual support, alleviating information loss caused by limited contextual summarization and improving representation quality under heterogeneous evaluation settings. Extensive experiments and ablation studies were conducted on text classification and named entity recognition tasks across multiple public benchmark datasets, using GPT-2 Small, GPT-2 Medium, and LLaMA3-8B as backbone models. The results demonstrate that the proposed framework consistently outperforms several mainstream fine-tuning methods in terms of accuracy, F1 score, and robustness, while also showing stable behavior under learning-rate sensitivity and missing-information settings. Overall, this study provides a novel perspective for efficient and interpretable fine-tuning paradigms, achieving a favorable balance among performance improvement, parameter efficiency, and deployment feasibility, and offering a practical basis for future extensions to more complex downstream scenarios.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- Arizona State University, Tempe, Arizona, United States of America.
- University of Pennsylvania, Philadelphia, Pennsylvania, United States of America.
- University of Southern California, Los Angeles, California, United States of America.
- Texas A&M University, College Station, Texas, United States of America.
- Georgia Institute of Technology, Atlanta, Georgia, United States of America.
- Trine University, Phoenix, Arizona, United States of America.
ORCID-Profile: Shuaidong Pan
Zitieren
APA
Ma, Y., Xing, Y., Wu, D., Zhou, Y., Zi, Y., Wang, M., Deng, Y., & Pan, S. (2026). Research on fine-tuning algorithms for Large Language Models integrating Uncertainty Modeling and External Memory Augmentation. PLOS One. https://doi.org/10.1371/journal.pone.0351493
Vancouver
Ma Y, Xing Y, Wu D, Zhou Y, Zi Y, Wang M, et al. Research on fine-tuning algorithms for Large Language Models integrating Uncertainty Modeling and External Memory Augmentation. PLOS One. 2026. doi:10.1371/journal.pone.0351493
BibTeX
@article{Ma2026,
author = {Ma, Yumeng and Xing, Yue and Wu, Di and Zhou, Yining and Zi, Yun and Wang, Ming and Deng, Yingnan and Pan, Shuaidong},
title = {Research on fine-tuning algorithms for Large Language Models integrating Uncertainty Modeling and External Memory Augmentation},
journal = {PLOS One},
year = {2026},
doi = {10.1371/journal.pone.0351493}
}
EUVIMED