Cross-Modal Temporal Attention Network for Continuous Authentication in Mobile Behavioral Biometrics Mobil Davranişsal Biyometride Sürekli Kimlik Doǧrulama için Çapraz Modal Zamansal Dikkat Aǧi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11636594
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Behavioral biometrics, continuous authentication, deep learning, mobile sensor data, multi-modality
- Atatürk Üniversitesi Adresli: Evet
Özet
In this study, a novel deep learning architecture called Cross-Modal Temporal Attention Network (CMTA-Net) is proposed for the problem of continuous authentication based on mobile behavioral biometrics. The model represents user behavior by processing touch, accelerometer, and gyroscope data through separate branches using cross-modal attention and temporal attention mechanisms. In experiments conducted on the HMOG dataset under the leakage-free Leave-One-Session-Out (LOSO) protocol, an EER of 8.17% was achieved at the instant decision level, which was reduced to 5.97% with 20-second decision fusion. The results demonstrate that the proposed approach provides effective and competitive performance for continuous authentication using multimodal mobile sensor data.