.png){kind=icon}
.png){kind=icon}
Return
KANFormer: A Kolmogorov-Arnold Network architecture for accurate and interpretable vehicle emission prediction
.png){kind=icon}
PRE
AI
X
R
X
S
DOI：10.1016/j.apr.2026.103109.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A KANFormer model is proposed for accurate real-world vehicle emission forecasting. • Kolmogorov–Arnold Networks capture nonlinear emission dynamics effectively. • The positioning of Transformer layers in KAN-based architectures influences prediction performance. • Strong performance is achieved with limited training data from a small number of drivers. • Spline-based analysis provides interpretable insights into driving behavior-emission relationships.
Journal
IF:
3.5
Papers:
3.0K
Citations:
7.4K
Organization
Cited Papers