Zobrazeno 1 - 10
of 637
pro vyhledávání: '"Wei Zhipeng"'
Jailbreaking attacks show how Large Language Models (LLMs) can be tricked into generating harmful outputs using malicious prompts. To prevent these attacks, other LLMs are often used as judges to evaluate the harmfulness of the generated content. How
Externí odkaz:
http://arxiv.org/abs/2411.01077
Autor:
Gao, Ziyi, Chen, Kai, Wei, Zhipeng, Mou, Tingshu, Chen, Jingjing, Tan, Zhiyu, Li, Hao, Jiang, Yu-Gang
Recent diffusion-based unrestricted attacks generate imperceptible adversarial examples with high transferability compared to previous unrestricted attacks and restricted attacks. However, existing works on diffusion-based unrestricted attacks are mo
Externí odkaz:
http://arxiv.org/abs/2408.05479
Text-to-image models encounter safety issues, including concerns related to copyright and Not-Safe-For-Work (NSFW) content. Despite several methods have been proposed for erasing inappropriate concepts from diffusion models, they often exhibit incomp
Externí odkaz:
http://arxiv.org/abs/2407.12383
Publikováno v:
Frontiers in Earth Science, Vol 9 (2021)
For economic and efficient development of extremely high-condensate shale gas reservoirs, a numerical model of segmental multicluster fractured horizontal well was established considering the effect of condensate and desorption, and the optimization
Externí odkaz:
https://doaj.org/article/3d62a3719f3d416eb5b472c3f16ed75b
Investigation of localized and delocalized excitons in ZnO/ZnS core-shell heterostructured nanowires
Autor:
Li Ruxue, Wei Zhipeng, Zhao Fenghuan, Gao Xian, Fang Xuan, Li Yongfeng, Wang Xinwei, Tang Jilong, Fang Dan, Wang Haizhu, Chen Rui, Wang Xiaohua
Publikováno v:
Nanophotonics, Vol 6, Iss 5, Pp 1093-1100 (2016)
The localized states in ZnO nanowires (NWs) through the growth of ZnS shell have been introduced in this paper. Morphology and optical properties of the ZnO/ZnS core-shell heterostructured NWs after different rapid thermal annealing (RTA) treatments
Externí odkaz:
https://doaj.org/article/64d1679847064df3a909a10e45a886be
In this paper, we propose a novel transfer-based targeted attack method that optimizes the adversarial perturbations without any extra training efforts for auxiliary networks on training data. Our new attack method is proposed based on the observatio
Externí odkaz:
http://arxiv.org/abs/2209.03716
Publikováno v:
In Cell Reports Physical Science 16 October 2024 5(10)
Autor:
Kang, Yubin, Meng, Bingheng, Hou, Xiaobing, Wang, Puning, Tang, Jilong, Wang, Le, Wei, Zhipeng
Publikováno v:
In Vacuum October 2024 228
Publikováno v:
In Optics and Laser Technology September 2024 176
Recent studies have shown that adversarial examples hand-crafted on one white-box model can be used to attack other black-box models. Such cross-model transferability makes it feasible to perform black-box attacks, which has raised security concerns
Externí odkaz:
http://arxiv.org/abs/2112.05379