Výsledky vyhledávání

Report

Test-time Adversarial Defense with Opposite Adversarial Path and High Attack Time Cost

Autor: Yeh, Cheng-Han, Yu, Kuanchun, Lu, Chun-Shien

Deep learning models are known to be vulnerable to adversarial attacks by injecting sophisticated designed perturbations to input data. Training-time defenses still exhibit a significant performance gap between natural accuracy and robust accuracy. I

Externí odkaz: http://arxiv.org/abs/2410.16805

Zobrazit plný text záznamu

Report

Adversarial Robustness Overestimation and Instability in TRADES

Autor: Li, Jonathan Weiping, Liang, Ren-Wei, Yeh, Cheng-Han, Tsai, Cheng-Chang, Yu, Kuanchun, Lu, Chun-Shien, Chen, Shang-Tse

This paper examines the phenomenon of probabilistic robustness overestimation in TRADES, a prominent adversarial training method. Our study reveals that TRADES sometimes yields disproportionately high PGD validation accuracy compared to the AutoAttac

Externí odkaz: http://arxiv.org/abs/2410.07675

Zobrazit plný text záznamu

Report

Defending Against Repetitive Backdoor Attacks on Semi-supervised Learning through Lens of Rate-Distortion-Perception Trade-off

Autor: Lee, Cheng-Yi, Kao, Ching-Chia, Yeh, Cheng-Han, Lu, Chun-Shien, Yu, Chia-Mu, Chen, Chu-Song

Semi-supervised learning (SSL) has achieved remarkable performance with a small fraction of labeled data by leveraging vast amounts of unlabeled data from the Internet. However, this large pool of untrusted data is extremely vulnerable to data poison

Externí odkaz: http://arxiv.org/abs/2407.10180

Zobrazit plný text záznamu

Report

CIMulator: A Comprehensive Simulation Platform for Computing-In-Memory Circuit Macros with Low Bit-Width and Real Memory Materials

Autor: Le, Hoang-Hiep, Baig, Md. Aftab, Hong, Wei-Chen, Tsai, Cheng-Hsien, Yeh, Cheng-Jui, Liang, Fu-Xiang, Huang, I-Ting, Tsai, Wei-Tzu, Cheng, Ting-Yin, De, Sourav, Chen, Nan-Yow, Lee, Wen-Jay, Lin, Ing-Chao, Chang, Da-Wei, Lu, Darsen D.

This paper presents a simulation platform, namely CIMulator, for quantifying the efficacy of various synaptic devices in neuromorphic accelerators for different neural network architectures. Nonvolatile memory devices, such as resistive random-access

Externí odkaz: http://arxiv.org/abs/2306.14649

Zobrazit plný text záznamu

Akademický článek

Suppressing Interface Defects in Perovskite Solar Cells via Introducing a Plant-Derived Ergothioneine Self-Assembled Monolayer.

Autor: Yeh, Cheng-Hsien^1,2 (AUTHOR) n28084012@mail.ncku.edu.tw, Hsu, Hung-Chieh^1,3 (AUTHOR) n28084020@gs.ncku.edu.tw, Tsao, Jung-Che^1,3 (AUTHOR) jerry881015@gmail.com, Wu, Hsuan-Ta⁴ (AUTHOR) htwu@must.edu.tw, Lin, Teh-Pei⁵ (AUTHOR) drtplin@gmail.com, Wu, Chien-Te⁵ (AUTHOR) ctwu168@gmail.com, Wu, Shih-Hsiung³ (AUTHOR) shihhsiung@itri.org.tw, Shih, Chuan-Feng^1,2,6 (AUTHOR) shihhsiung@itri.org.tw

Publikováno v: Materials (1996-1944). Dec2024, Vol. 17 Issue 23, p5739. 14p.

Zobrazit plný text záznamu

Plný text ve formátu HTML

Vyhledávací nástroje:

Upřesnit hledání