CORTEXA
← Browse

Xiaoxiao Miao

1 paper indexed

arxivcs.CV2026-07-16

ARMOR++: Agentic Orchestration of a Multi-Domain Primitive Set for Transferable Attacks on Deepfake Detectors

Christos Korgialas, Gabriel Lee Jun Rong, Dion Jia Xu Ho, Pai Chet Ng, Xiaoxiao Miao, Konstantinos N. Plataniotis

The reliability of deepfake detectors frequently degrades under black-box adversarial transfer, as these models often rely on fragile, architecture-dependent forensic cues. Existing transfer attacks often lack semantic awareness and struggle to maintain effectiveness under strict…

View free PDFSource page