TADA adapts steganalysis models to unknown JPEG processing pipelines via data emulation from small unlabeled sets, yielding gains in robustness to cover source mismatch over baselines.
Title resolution pending
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
verdicts
UNVERDICTED 4representative citing papers
A Bayesian optimal experimental design framework with Gaussian approximation of expected information gain and surrogate Fisher information enables optimized uniaxial tests that significantly improve identifiability of history-dependent constitutive parameters over random designs.
A self-supervised Degradation Estimation Network estimates parameters for physics-informed noise distributions to generate realistic synthetic low-light data, showing gains on noise replication, enhancement, and detection tasks.
PixIE proposes a feed-forward pixel-space low-light image enhancement network using DINO-prompted pixel blocks, spatial-channel compaction, and multi-receptive-field embeddings, claiming 1.9-15.0% PSNR gains and 8.5-44.4% LPIPS reductions on benchmarks.
citing papers explorer
No citing papers match the current filters.