Intermediate layer embedding sensitivity to perturbations distinguishes AI-generated images from real ones, yielding higher AUROC on GenImage and Forensics Small benchmarks than prior methods.
Dimensionality compression and expansion in deep neural networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Intermediate Representations are Strong AI-Generated Image Detectors
Intermediate layer embedding sensitivity to perturbations distinguishes AI-generated images from real ones, yielding higher AUROC on GenImage and Forensics Small benchmarks than prior methods.