14 datasets found

The dataset used in the paper is a collection of images with perturbations, where participants are asked to identify when an image becomes just noticeably different from the...

The dataset used in the paper to evaluate the harmlessness alignment of multimodal large language models (MLLMs). The dataset consists of 750 harmful instructions paired with...

The dataset used in the paper to evaluate the effectiveness of the BEEAR method in mitigating safety backdoors in instruction-tuned LLMs.

The Labeled Faces in the Wild (LFW) dataset contains 13,233 images from 5,749 identities, with large variations in pose, expression and illumination.

BLUFF is an interactive system for visualizing, characterizing, and deciphering adversarial attacks on DNNs.

The VBCDE-100 dataset is a comprehensive dataset containing 100 sensitive prompts spread across five categories: violence, gore, illegal activities, discrimination, and...

Randomized neural networks are vulnerable to adversarial attacks. We investigate the effect of proxy-gradient-based attacks on randomized neural networks and propose a...

Torchattacks is a PyTorch library that contains adversarial attacks to generate adversarial examples and to verify the robustness of deep learning models.

Speaker identification models are vulnerable to carefully designed adversarial perturbations of their input signals that induce misclas-sification.

LocalStyleFool: Regional Video Style Transfer Attack Using Segment Anything

The dataset used in the paper to test the Gradient Cuff method for detecting jailbreak attacks on large language models.

The dataset used in this paper is a perturbation benchmark containing nineteen common corruptions and five adversarial attacks.

Counterfactual explanations and adversarial attacks have a related goal: flipping output labels with minimal perturbations regardless of their characteristics.

A defense framework for segmentation models against adversarial attacks in medical imaging