arXiv:2608. 06674v1 Announce Type: cross Abstract: Adversarial vulnerabilities remain a major concern for the safe deployment of neural networks, particularly in object detection, a core task embedded in many safety-critical systems.
arXiv:2606. 03879v1 Announce Type: cross Abstract: As foundation models scale toward fusing more heterogeneous visual streams, understanding how diverse encoders interact under joint training becomes a prerequisite for principled design.
arXiv:2606.22516v2 Announce Type: replace Abstract: Input Diversity (DI), a random resize and pad applied at each attack iteration, is a near-default ingredient of transfer-based attacks, widely assu...
arXiv:2604. 21395v3 Announce Type: replace-cross Abstract: Ordinary supervised training minimises the task loss and then stops.
arXiv:2505. 03646v5 Announce Type: replace-cross Abstract: Adversarial robustness of deep autoencoders (AEs) has received less attention than that of discriminative models, although their compressed latent representations induce ill-conditioned mappings that can amplify small input perturbations and destabilize reconstructions.
arXiv:2607. 26574v2 Announce Type: replace-cross Abstract: Safety classifiers ("guards") are the dominant black-box defense for vision-language models, yet a guard judges an input's surface form, not its meaning: a harmful request re-encoded as set theory, formal logic, a classical language, code, or text rendered inside an image slips past a guard that would block it in plain language - the decode gap.
The paper presents a preprocessor that recovers and decodes encoded content in vision‑language models to close the decode gap that allows harmful requests to bypass safety classifiers. Evaluated against eleven encoding attacks, the preprocessor raises block rates from 0 % to 67‑90 % but also increases benign over‑refusal, and no configuration achieves an ensemble attack‑success rate below 40 % while keeping benign over‑refusal under 70 %. The study shows that closing one encoding channel merely relocates success rather than eliminating it, highlighting the limits of recovery‑based defenses.
The paper introduces Median Temporal Ensembling, a training‑free aggregation method for action‑chunked visuomotor policies that replaces the standard exponential weighted mean with a coordinate‑wise median. This approach remains robust against adversarial corruption, maintaining a high recovery rate even as attack strength increases, and performs at least as well as the mean across numerous configurations while improving in many cases. It also handles non‑adversarial failures such as blank camera frames and shows limited impact on clean data, though it cannot counteract uniform shifts applied to all predictions.
The paper evaluates six preprocessing defenses against adversarial attacks on depthwise‑separable CNNs, the dominant architecture in edge vision systems, and finds that these defenses consistently fail to recover clean predictions for such models, whereas a residual architecture shows partial recovery. The study reveals that the same preprocessing steps that break clean predictions leave adversarial predictions largely intact, creating a measurable asymmetry that can be exploited for detection without retraining or architectural changes. It also demonstrates that common image quality metrics do not reliably indicate defense effectiveness, highlighting a methodological gap in current evaluation practices.
arXiv:2609.39134v1 Announce Type: new Abstract: Visual-token compression improves the efficiency of large vision-language models, but can expose failures that full-token evaluation misses. We study a...
The paper examines how preprocessing defenses, commonly used to protect edge vision systems, perform on depthwise‑separable CNNs versus residual architectures. Six preprocessing methods were tested against adversarial attacks, revealing that depthwise‑separable models consistently fail to recover from perturbations while residual models show partial recovery. Interestingly, the same preprocessing that hinders clean predictions leaves adversarial predictions largely intact, offering a measurable detection signal, and the study also finds that typical image‑quality metrics do not reliably indicate defense success.
arXiv:2607. 26574v1 Announce Type: cross Abstract: Safety classifiers ("guards") are the dominant black-box defense for vision-language models, yet they judge an input's surface form, not its meaning: a harmful request re-encoded as set theory, formal logic, a rare language, code, or an image of text slips past a guard that would block it in plain language -- the decode gap.