Evaluating Kolmogorov-Arnold Networks for Spatio-temporal Mobility Based User Identification
2025 16th International Conference on Electrical and Electronics Engineering, ELECO 2025, İstanbul, Türkiye, 27 - 29 Kasım 2025, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/eleco69582.2025.11329214
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Ankara Yıldırım Beyazıt Üniversitesi Adresli: Evet
Özet
Spatio-temporal fingerprinting plays a crucial role in mobility analysis, where accurate user reidentification requires models capable of handling both spatial and temporal dependencies. It has been shown that by integrating spatial map structure, temporal dynamics, and movement statistics, it is possible to achieve remarkable trajectory fingerprinting performance. Building upon this foundation, the present study investigates the effectiveness of Kolmogorov-Arnold Networks (KANs) compared to conventional Multilayer Perceptrons (MLPs) for sequential spatio-temporal fingerprinting. Experimental results reveal that KANs consistently outperform MLPs, providing an improvement of 7% in accuracy. These findings demonstrate the potential of KAN-based architectures to advance fingerprinting performance, offering higher accuracy, stronger generalization, and enhanced robustness for sequential spatio-temporal data analysis.