This paper devises a backdoor attack detection method to identify whether a sample is attacked for image-related applications. The information consistence ...
A num- ber of backdoor attack detection methods are proposed to iden- tify whether an image is contaminated. However, the current methods may not efficiently ...
TL;DR: In this paper, a backdoor attack detection method is proposed to identify whether a sample is attacked for image-related applications, ...
Jul 19, 2021 · This method is based on detection of early diagnosis of skin cancer cells. The experimental result shows that the proposed algorithm gives ...
Oct 23, 2024 · We propose several properties for an effective and feasible backdoor attack on large pre-trained models and introduce a new threat model based ...
In this paper, we proposed a novel backdoor attack on semantic segmentation named object-free backdoor attack (OFBA).
In this work, we explore backdoor attacks on segmentation models to misclassify all pixels of a victim class by injecting a specific trigger on non-victim ...
People also ask
What are some measures that could be taken to prevent a backdoor attack?
What is a real life example of a backdoor attack?
What is the most common backdoor attack?
How does a backdoor attack work?
Based on that, we propose an effective, generalizable approach based on channel shuffling to detect backdoor attacks and recover triggers with minimal overhead.
This paper proposes a novel attack paradigm, the fine-grained attack, where the target label is treated from the object-level instead of the image-level.
We developed a novel technique that automatically generates alternate triggers to increase the effectiveness of triggers.