Zobrazeno 1 - 10
of 114
pro vyhledávání: '"Gai Kuo"'
Deep neural networks (DNNs) are vulnerable to small adversarial perturbations of the inputs, posing a significant challenge to their reliability and robustness. Empirical methods such as adversarial training can defend against particular attacks but
Externí odkaz:
http://arxiv.org/abs/2408.00329
Autor:
Gai, Kuo, Zhang, Shihua
In practice, deeper networks tend to be more powerful than shallow ones, but this has not been understood theoretically. In this paper, we find the analytical solution of a three-layer network with a matrix exponential activation function, i.e., $$ f
Externí odkaz:
http://arxiv.org/abs/2407.02540
Neural collapse (NC) is a simple and symmetric phenomenon for deep neural networks (DNNs) at the terminal phase of training, where the last-layer features collapse to their class means and form a simplex equiangular tight frame aligning with the clas
Externí odkaz:
http://arxiv.org/abs/2405.00985
Publikováno v:
In Pattern Recognition October 2024 154
Autor:
Gai, Kuo, Zhang, Tongrui, Xu, Zhengyi, Li, Guangzhao, He, Zihan, Meng, Shuhuai, Shi, Yixin, Zhang, Yuheng, Zhu, Zhou, Pei, Xibo, Wang, Jian, Wan, Qianbing, Cai, He, Li, Yijun, Chen, Junyu
Publikováno v:
In Chemical Engineering Journal 1 August 2024 493
Autor:
Gai, Kuo, Zhang, Shihua
Recent studies revealed the mathematical connection of deep neural network (DNN) and dynamic system. However, the fundamental principle of DNN has not been fully characterized with dynamic system in terms of optimization and generalization. To this e
Externí odkaz:
http://arxiv.org/abs/2102.09235
Autor:
Gai, Kuo, Zhang, Shihua
Non-adversarial generative models such as variational auto-encoder (VAE), Wasserstein auto-encoders with maximum mean discrepancy (WAE-MMD), sliced-Wasserstein auto-encoder (SWAE) are relatively easy to train and have less mode collapse compared to W
Externí odkaz:
http://arxiv.org/abs/2005.09923
Principal component analysis (PCA) is one of the most widely used dimension reduction and multivariate statistical techniques. From a probabilistic perspective, PCA seeks a low-dimensional representation of data in the presence of independent identic
Externí odkaz:
http://arxiv.org/abs/1911.10796
Autor:
Chen, Junyu, Song, Li, Qi, Fangwei, Qin, Siyu, Yang, Xiangjun, Xie, Wenjia, Gai, Kuo, Han, Ying, Zhang, Xin, Zhu, Zhou, Cai, He, Pei, Xibo, Wan, Qianbing, Chen, Ning, Wang, Jian, Wang, Qi, Li, Yijun
Publikováno v:
In Nano Energy February 2023 106
Autor:
Chen, Jie, Liao, Lijun, Lan, Tingting, Zhang, Zhijun, Gai, Kuo, Huang, Yibing, Chen, Jinlong, Tian, Weidong, Guo, Weihua
Publikováno v:
In Applied Materials Today September 2020 20