Adversarial perturbations can alter the predictions of frozen vision-language models (VLMs) while leaving their confidence and image--text similarity patterns seemingly plausible. We investigate whether we can identify adversarial inputs based on the broader way an image interacts with a collection of general semantic...
Adversarial example detectors are often tied to the classifier backbone they were trained on, limiting reuse when the protected model is replaced or upgraded. Directly transferring such detectors across backbones is challenging because different networks generally produce incompatible internal representations. We propo...
Arash Vashagh, R. Razavi-Far· arXiv (Cornell University)· 0 citations
FedNIA injects random noise inputs to analyze the layerwise activation patterns in client models leveraging an autoencoder that detects abnormal behaviors indicative of data poisoning and can defend against diverse attack types, including sample poisoning, label flipping, and backdoors.
Ehsan Hallaji, R. Razavi-Far, Mehrdad Saif· IEEE Transactions on Knowled...· 2 citations
This paper investigates the certified and empirical robustness of various Kolmogorov-Arnold network architectures against strong evasion attacks and systematically evaluates the robustness of various defended and undefended KAN models under FGSM, PGD, and C&W attacks.
This perspective explores how learning systems may progress beyond reactive adaptation toward autonomous recognition, policy selection, and context-sensitive regulation of learning behavior under persistent uncertainty and positions SAL as a conceptual systems framework for organizing future research on resilient, long...
Ehsan Hallaji, R. Razavi-Far· Machine Learning and Knowled...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
