Learning the manifold of authenticity : hybrid-curvature representation learning for generalizable deepfake detection

Larue, Nicolas (avtor), Štruc, Vitomir (avtor), Peer, Peter (avtor), Vu, Ngoc-Son (avtor)

The practical utility of deepfake detectors is crippled by a crisis of generalization: models that perform well on known manipulation techniques consistently fail when faced with unseen forgeries.We argue this failure stems from a fundamental geometric mismatch. Existing methods implicitly assume that the manifold of authentic faces can be modeled in a space of uniform curvature, typically Euclidean, which inade-quately captures the complex, multi-scale structure of facial features. This paper validates the hypothesis that authentic faces lie on a manifold whose geometry is inherently hybrid, requiring both angular compactness (a spherical property) and hierarchical organization (a hyperbolic property). To resolve this geometric mismatch, we introduce a novel detector, CTrue, that learns a unified, hybrid-curvature representation of facial authenticity. Trained exclusively on real faces via self-supervised learning, our method simultaneously projects facial embeddings onto two complementary manifolds: a hypersphere to enforce compactness and a hyperbolic space to model the natural feature hierarchy. A single set of mathematically-optimal prototypes acts as a ‘‘geometric bridge’’, unifying the learning objectives in both spaces. At inference, a composite score measures an embedding’s deviation from this learned manifold. On challenging cross-dataset and cross-manipulation benchmarks, our method achieves competitive generalization under a strictly pristine-only training setting, showing that hybrid-curvature representations provide an effective and data-efficient alternative for deepfake detection.

• Jezik: Angleški jezik
• Ključne besede: contrast learning, generalized deepfake detection, hyperbolic, one-class anomaly detection
• Vrsta gradiva: Članek v reviji
• Tipologija: 1.01 - Izvirni znanstveni članek
• Organizacija: FRI - Fakulteta za računalništvo in informatiko; FE - Fakulteta za elektrotehniko
• Status publikacije: Objavljeno
• Različica publikacije: Objavljena publikacija
• Leto izida: 2026
• Št. strani: Str. 1-14
• Številčenje: Vol.
• PID: 20.500.12556/RUL-183410
• UDK: 004.932.2:004.8
• ISSN pri članku: 2169-3536
• DOI: 10.1109/ACCESS.2026.3702429
• COBISS.SI-ID: 281381123
• Datum objave v RUL: 12.06.2026

Gradivo je del revije

• Naslov: IEEE access
• Založnik: Institute of Electrical and Electronics Engineers
• ISSN: 2169-3536
• COBISS.SI-ID: 519839513

Licence

• Licenca: CC BY 4.0, Creative Commons Priznanje avtorstva 4.0 Mednarodna
• Povezava: http://creativecommons.org/licenses/by/4.0/deed.sl
• Opis: To je standardna licenca Creative Commons, ki daje uporabnikom največ možnosti za nadaljnjo uporabo dela, pri čemer morajo navesti avtorja.

Sekundarni jezik

• Jezik: Slovenski jezik
• Ključne besede: kontrastno učenje, posplošena detekcija globokih ponaredkov, hiperboličnost, eno-razredna detekcija anomalij