scholar.google.com › citations
Feb 1, 2024 · This study presents, for the first time, an exhaustive review of the transferability aspect of adversarial attacks.
Feb 26, 2024 · To study transferable adversarial attacks, the selection of a suitable baseline method is essential for benchmarking attack techniques. In this ...
Feb 16, 2024 · The core principle of the DI-FGSM is to introduce input diversity in the process of generating adversarial samples to find Semantic Similarity.
People also ask
What are the different types of adversarial attacks?
What is attack transferability?
What is an example of an adversarial attack in real life?
What is an example of adversarial targeting?
In this paper, we propose an effective adversarial example generator TopicAttack to disturb the inference of a target victim topic model.
Feb 9, 2024 · Zhibo Jin, Jiayu Zhang, Zhiyu Zhu, Huaming Chen: Benchmarking Transferable Adversarial Attacks. CoRR abs/2402.00418 (2024).
Transfer-based Black-box Attacks: Transfer-based at- tacks craft adversarial examples against a substitute model, which are probable to fool black-box ...
This paper describes a new method called feature-momentum adversarial attack (FMAA) to improve transferability.
Universal and Transferable Adversarial Attacks on Aligned Language Models, Greedy Coordinate Gradient (GCG), White-box access, Suffix attack, 256k queries, 256K ...
BlackboxBench is a comprehensive benchmark containing mainstream adversarial black-box attack methods implemented based on PyTorch.
Apr 14, 2024 · The study introduces the AdvBench benchmark, a tool for systematically assessing how well adversarial attacks can induce LLMs to produce harmful ...