Výsledky vyhledávání

Report

Autor: Choi, Ching Lam, Duplessis, Alexandre, Belongie, Serge

Gradient-based interpretations often require an anchor point of comparison to avoid saturation in computing feature importance. We show that current baselines defined using static functions--constant mapping, averaging or blurring--inject harmful col

Externí odkaz: http://arxiv.org/abs/2410.08069

Zobrazit plný text záznamu

Report

Universal Adversarial Directions

Autor: Choi, Ching Lam, Farnia, Farzan

Despite their great success in image recognition tasks, deep neural networks (DNNs) have been observed to be susceptible to universal adversarial perturbations (UAPs) which perturb all input samples with a single perturbation vector. However, UAPs of

Externí odkaz: http://arxiv.org/abs/2210.15997

Zobrazit plný text záznamu

Report

Self-distillation with Batch Knowledge Ensembling Improves ImageNet Classification

Autor: Ge, Yixiao, Zhang, Xiao, Choi, Ching Lam, Cheung, Ka Chun, Zhao, Peipei, Zhu, Feng, Wang, Xiaogang, Zhao, Rui, Li, Hongsheng

The recent studies of knowledge distillation have discovered that ensembling the "dark knowledge" from multiple teachers or students contributes to creating better soft targets for training, but at the cost of significantly more computations and/or p

Externí odkaz: http://arxiv.org/abs/2104.13298

Zobrazit plný text záznamu

Report

DivCo: Diverse Conditional Image Synthesis via Contrastive Generative Adversarial Network

Autor: Liu, Rui, Ge, Yixiao, Choi, Ching Lam, Wang, Xiaogang, Li, Hongsheng

Conditional generative adversarial networks (cGANs) target at synthesizing diverse images given the input conditions and latent codes, but unfortunately, they usually suffer from the issue of mode collapse. To solve this issue, previous works mainly

Externí odkaz: http://arxiv.org/abs/2103.07893

Zobrazit plný text záznamu