原创 [論文筆記] Universal Adversarial Perturbations Against Semantic Image Segmentation(ICCV 2017)

Universal Adversarial Perturbations Against Semantic Image Segmentation(ICCV 2017) 文章簡介: DataSet: Cityscapes 區別:

原创 [論文筆記]Universal adversarial perturbations(CVPR 2017)

Universal adversarial perturbations(CVPR 2017) 文章簡介: 本文主要是介紹了一種universal的擾動,能讓大部分圖片加入該噪聲後就能被誤分類,擾亂一個新的數據點只需要向圖像添加一

原创 [論文筆記]UNDERSTANDING AND ENHANCING THE TRANSFERABILITY OF ADVERSARIAL EXAMPLES(archive)

UNDERSTANDING AND ENHANCING THE TRANSFERABILITY OF ADVERSARIAL EXAMPLES(archive) 文章簡介 在本研究中,作者系統地研究了兩類可能影響對抗性例子遷移

原创 [論文筆記]Curls & Whey: Boosting Black-Box Adversarial Attacks(CVPR 2019 Oral)

Curls & Whey: Boosting Black-Box Adversarial Attacks(CVPR 2019 Oral) 文章簡介: 作者提出一種全新的black-box攻擊方法Curls&Whey,該方法可以

原创 [論文筆記]Rob-GAN: Generator, Discriminator, and Adversarial Attacker(CVPR 2019)

Rob-GAN: Generator, Discriminator, and Adversarial Attacker(CVPR 2019) 文章簡介: 本文主要是將adversarial attack加入到GAN的訓練過程中,

原创 [論文筆記]AttGAN: Facial Attribute Editing by Only Changing What You Want(2017 CVPR)

AttGAN: Facial Attribute Editing by Only Changing What You Want(2017 CVPR) 文章簡介: 本文研究面部屬性編輯任務,其目的是通過操作單個或多個感興趣的屬性(如

原创 [論文筆記]Fader Networks: Manipulating Images by Sliding Attributes(2017 NIPS)

Fader Networks: Manipulating Images by Sliding Attributes(2017 NIPS) 文章簡介: 本文介紹了一種新的encoder-decoder結構,該結構通過訓練,將圖像的顯

原创 [論文閱讀筆記]Trust Region Based Adversarial Attack on Neural Networks

Trust Region Based Adversarial Attack on Neural Networks(2019 CVPR) 文章簡介: Method: 本文主要採用Trust Region(信賴域) 優化算法, 該方

原创 [論文閱讀筆記]Towards Evaluating the Robustness of Neural Networks(CW)

Towards Evaluating the Robustness of Neural Networks(C&W)(2017 Best Student Paper) 文章簡介: 證明defensive distillation不

原创 [論文閱讀筆記] Adversarial Examples Are Not Bugs, They Are Features

Adversarial Examples Are Not Bugs, They Are Features(CVPR 2019) 文章核心: 作者認爲,對抗性樣本的存在不是網絡架構的問題,而是數據集的一個屬性。(non−robust

原创 [論文閱讀筆記] Boosting Adversarial Attacks with Momentum

Boosting Adversarial Attacks with Momentum(CVPR2018) Source Code: https://github.com/dongyp13/Non-Targeted-Adversar

原创 [論文閱讀筆記]SEMANTIC ADVERSARIAL ATTACKS: PARAMETRIC TRANSFORMATIONS THAT FOOL DEEP CLASSIFIERS(ICCV)

SEMANTIC ADVERSARIAL ATTACKS: PARAMETRIC TRANSFORMATIONS THAT FOOL DEEP CLASSIFIERS(2019 ICCV) 文章簡介: 本文的出發點不同於以前的方法

原创 [論文閱讀筆記]Adversarial Examples that Fool both Computer Vision and Time-Limited Humans

Adversarial Examples that Fool both Computer Vision and Time-Limited Humans(CVPR2018) 文中提出了首個可以欺騙人類的對抗樣本。左圖是貓右圖是對抗樣

原创 [論文閱讀筆記]The Limitations of Deep Learning in Adversarial Settings

The Limitations of Deep Learning in Adversarial Settings 參考 知乎 CSDN 文章概述: 與之前的基於提高原始類別標記的損失函數或者降低目標類別標記的損失函數的方式不同,這

原创 [論文閱讀筆記]Adversarial Transformation Networks: Learning to Generate Adversarial Examples

Adversarial Transformation Networks: Learning to Generate Adversarial Examples 在現有的許多方法中,利用梯度信息進行攻擊的方法佔絕大多數,本文另闢蹊徑,