EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Quantifying Epistemic Uncertainty in Multimodal Long-Tailed Classification: A Belief Entropy-Based Evidential Fusion Framework

Guorui Zhu iD

Entropy · 19.03.2026 · 28 (3) · 343

DOI: 10.3390/e28030343 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Entropy, online erschienen am 19.03.2026. Verfasst von Guorui Zhu. Beteiligte Einrichtung: School of Business Administration, Southwestern University of Finance and Economics, Chengdu 611130, China. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep multimodal learning has excelled in tasks involving vision, language, and audio modalities. Nevertheless, their performance on tail classes exhibits significant degradation under the long-tailed distributions common in real-world data, meanwhile related fusion schemes often provide only limited treatment of modality-specific uncertainty and rarely incorporate explicit mechanisms for class-level fairness. To address these information discrepancies, we present a framework that integrates evidential reasoning with deep learning–Uncertainty-Quantified Multimodal Learning for Long-Tailed Classification (UMuLT). The framework includes: (i) an uncertainty-gated evidential fusion module that adaptively down-weights unreliable modalities; (ii) an exponential moving average (EMA) fairness regularizer that dynamically amplifies tail-class gradients; and (iii) a cross-modal consistency regularizer optimized in two stages: tail specialization with lightweight adapters on tail-class data to obtain a balanced initialization, followed by end-to-end fine-tuning. The effectiveness and practicality of our method are verified on three long-tailed benchmarks for multimodal classification. Experiments show consistent gains over strong baselines in overall metrics, calibration, and tail subset performance. Statistical significance tests confirm the superiority of the proposed framework.

Schlagwörter

Stichwörter der Autor:innen

Dempster-Shafer Theorybelief entropylong-tailed classificationmultimodal information fusionuncertainty quantification

Autor:innen und Institutionen

  • School of Business Administration, Southwestern University of Finance and Economics, Chengdu 611130, China.
  • Institute of Big Data, Southwestern University of Finance and Economics, Chengdu 611130, China.

ORCID-Profile: Guorui Zhu

Zitieren

APA

Zhu, G. (2026). Quantifying Epistemic Uncertainty in Multimodal Long-Tailed Classification: A Belief Entropy-Based Evidential Fusion Framework. Entropy, 28(3), 343. https://doi.org/10.3390/e28030343

Vancouver

Zhu G. Quantifying Epistemic Uncertainty in Multimodal Long-Tailed Classification: A Belief Entropy-Based Evidential Fusion Framework. Entropy. 2026;28(3):343. doi:10.3390/e28030343

BibTeX
@article{Zhu2026,
  author = {Zhu, Guorui},
  title = {Quantifying Epistemic Uncertainty in Multimodal Long-Tailed Classification: A Belief Entropy-Based Evidential Fusion Framework},
  journal = {Entropy},
  year = {2026},
  volume = {28},
  number = {3},
  pages = {343},
  doi = {10.3390/e28030343}
}
Kostenlos starten →