keyword

white-box ensemble

A white-box ensemble is a collection of multiple machine learning models whose internal architectures, parameters, and gradient information are fully accessible and transparent during optimization or evaluation. In adversarial machine learning, this configuration is commonly utilized to construct adversarial examples or physical patches by simultaneously optimizing the input across all constituent models. By aggregating the losses and gradients of diverse fully observable networks at the same time, the optimization avoids overfitting to the specific decision boundaries of a single classifier, thereby producing adversarial perturbations that exhibit significantly higher robustness, universality, and cross-model transferability against unseen target systems.

1 item

Adversarial Patch

Tom B. Brown, Dandelion Mané, Aurko Roy, Martín Abadi, Justin Gilmer

OrganizationsGoogle

Why you should read this

Introduces printable, scene-agnostic adversarial patches capable of overriding deep learning image classifiers in the physical world to force the prediction of an arbitrary target class.

We present a method to create universal, robust, targeted adversarial image patches in the real world. The patches are universal because they can be used to attack any scene, robust because they work under a wide variety of transformations, and targeted because they can cause a classifier to output any target class. These adversarial patches can be printed, added to any scene, photographed, and presented to image classifiers; even when the patches are small, they cause the classifiers to ignore the other items in the scene and report a chosen target class. To reproduce the results from the paper, our code is available at this https URL

Source

https://arxiv.org/pdf/1712.09665.pdf

Added

2026-09-25