SoK: How Robust is Image Classification Deep Neural Network Watermarking?
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Deep Neural Network (DNN) watermarking is a method for provenance verification of DNN models. Watermarking should be robust against watermark removal attacks that derive a surrogate model that evades provenance verification. Many watermarking schemes that claim robustness have been proposed, but their robustness is only validated in isolation against a relatively small set of attacks. There is no systematic, empirical evaluation of these claims against a common, comprehensive set of removal attacks. This uncertainty about a watermarking scheme’s robustness causes difficulty to trust their deployment in practice. In this paper, we evaluate whether recently proposed watermarking schemes that claim robustness are robust against a large set of removal attacks. We survey methods from the literature that (i) are known removal attacks, (ii) derive surrogate models but have not been evaluated as removal attacks, and (iii) novel removal attacks. Weight shifting and smooth retraining are novel removal attacks adapted to the DNN watermarking schemes surveyed in this paper. We propose taxonomies for watermarking schemes and removal attacks. Our empirical evaluation includes an ablation study over sets of parameters for each attack and watermarking scheme on the image classification datasets CIFAR-10 and ImageNet. Surprisingly, our study shows that none of the surveyed watermarking schemes is robust in practice. We find that schemes fail to withstand adaptive attacks and known methods for deriving surrogate models that have not been evaluated as removal attacks. This points to intrinsic flaws in how robustness is currently evaluated. Our evaluation includes a discussion of the runtime of each attack to underpin their practical relevance. While none of the schemes is robust against all attacks, none of the attacks removes all watermarks. We show that attacks can be combined and find combined attacks that remove all watermarks. We show that watermarking schemes need to be evaluated against a more extensive set of removal attacks with a more realistic adversary model. Our source code and a complete dataset of evaluation results are publicly available, which allows to independently verify our conclusions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it