Analysis of the Sensitivity Attack against Electronic Watermarks in Images

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Evading Classifiers by Morphing in the Dark (2017, background): “We believe that this scoring mechanism can be used to relax the assumption on the availability of real-value scores that other settings [20, 26, 35] make.”
Data-Hiding Codes (2005, unknown): “Sensitivity Attacks [124]–[126].”
The First 50 Years of Electronic Watermarking (2002, background): “Researchers have documented many different attacks that an adversary might apply, for example, collusion attacks [8, 33], ambiguity attacks [34], copy attacks [35], sensitivity and gradient descent attacks [12, 13].”
Spread-spectrum watermarking of audio signals (2003, background): “…the most significant deficiency of both schemes is that by breaking a single player (debugging, reverse engineering, or the sensitivity attack [12]), one can extract the secret information (the SS sequence or the hidden quantizers in QIM) and recreate the original (in the case of SS) or…”
Can machine learning be secure? (2006, background): “Randomization of thresholds within the watermark verification algorithm increase the number of probes necessary to remove a digital watermark [19].”