發(fā)簡(jiǎn)信
IP屬地：湖北
-
JPEG-resistant Adversarial Images
摘要： While JPEG compression is not differentiable, we show how to closely...
-
研讀筆記-COUNTERING ADVERSARIAL IMAGES USING INPUT TRANSFORMATIONS
文章鏈接：http://arxiv.org/pdf/1711.00117項(xiàng)目源碼：https://github.com/facebookrese...
-
研讀筆記-A study of the effect of JPG compression on adversarial images文章鏈接：http://arxiv.org/pdf/1608.00853 摘要：文中評(píng)估JPG壓縮對(duì)對(duì)抗性圖像分類的影響。對(duì)于小幅度的FGSM擾...
-
研讀筆記-Universal adversarial perturbations文章鏈接：https://arxiv.org/abs/1610.08401 摘要：提出一個(gè)系統(tǒng)的方法可以生成 universal perturb...
-
研讀筆記-improving the robustness of deep neural networks via stability training
Related work Data augmentation:提高模型識(shí)別能力的方法之一是增大樣本數(shù)據(jù)，往往更大的樣本數(shù)據(jù)集可以訓(xùn)練出表現(xiàn)更好的...