Reaching a Better Trade-Off Between Image Quality and Attack Success Rates in Transfer-Based Adversarial Attacks
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
We study transfer-based adversarial attacks that introduce perturbations in an image that are large enough to make an unknown CNN wrongly classify it. These perturbations should also be small enough to not be perceived by others. We denote this problem as mitigating the trade-off between image quality (or imperceptibility) and obtaining a high attack success rate (ASR). Our proposed methods explicitly integrate perceptual models into the attacks. We first propose a “spatial” perceptual-aware attack which allows the introduction of higher perturbations in the perceptually insignificant image regions, and less perturbations in the visually sensitive ones. We then propose a novel frequency-perceptual attack utilizing frequency perceptual models. Both types of attacks are independent of the gradient estimation, thus they can be directly incorporated into existing gradient-based attacks. Experiments demonstrate their effectiveness in mitigating such trade-offs.
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 | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| 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