EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Diffusion-Enhanced Temporal Transformer for Imbalanced Credit Card Fraud Detection

Mina Keller

International Journal of Business and Social Science Studies · 28.10.2026 · 6 (2) · 310-318

DOI: 10.66795/ijbsss64 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Business and Social Science Studies, online erschienen am 28.10.2026. Verfasst von Mina Keller. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Severe label imbalance and irregular transaction arrival times challenge fraud detection systems. We propose a diffusion-enhanced temporal Transformer (DETT) that generates complete fraud-positive transaction windows with a denoising diffusion model and classifies them with causal, interval-aware self-attention, keeping all synthesis confined to training data. On 3,600 simulated transaction windows with disjoint training, validation and test blocks, DETT achieves the highest mean F1 among all evaluated models across three independently seeded cohorts (0.9576 ± 0.0364), compared with 0.9238 ± 0.0421 for the non-augmented temporal Transformer and 0.9388 ± 0.0564 for matched random oversampling. A factorial ablation separates the contributions of window-level augmentation and interval-aware attention and shows that augmentation improves the operating point of both attention variants, while ranking quality remains high for all sequence models (mean PR-AUC 0.9859 for DETT). The results establish window-level diffusion as an effective, training-only minority-augmentation strategy for sequential fraud detection and motivate evaluation on account-aware real transaction histories.

Zitieren

APA

Keller, M. (2026). Diffusion-Enhanced Temporal Transformer for Imbalanced Credit Card Fraud Detection. International Journal of Business and Social Science Studies, 6(2), 310-318. https://doi.org/10.66795/ijbsss64

Vancouver

Keller M. Diffusion-Enhanced Temporal Transformer for Imbalanced Credit Card Fraud Detection. International Journal of Business and Social Science Studies. 2026;6(2):310-318. doi:10.66795/ijbsss64

BibTeX
@article{Keller2026,
  author = {Keller, Mina},
  title = {Diffusion-Enhanced Temporal Transformer for Imbalanced Credit Card Fraud Detection},
  journal = {International Journal of Business and Social Science Studies},
  year = {2026},
  volume = {6},
  number = {2},
  pages = {310--318},
  doi = {10.66795/ijbsss64}
}
Kostenlos starten →