Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks
論文簡介 在這項工作中，我們研究了一種新的攻擊類型，稱爲乾淨標籤攻擊，攻擊者注入的訓練示例被認證機構清晰地標記，而不是被攻擊者自己惡意地貼上標籤。我們的策略假設攻擊者不瞭解訓練數據，而是瞭解模型及其參數。攻擊者的目標是當網絡在包含中毒實例的增強數據集上進行重新訓練後，使重新訓練的網絡將一個特定測試實例從一個類錯誤地分類爲她選擇的另一個類。除了目標的預期預測錯誤之外，受害的分類器的性能下降並不明顯。
相關文章
- 1. 【翻譯】Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks
- 2. Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks 論文閱讀、復現及思考
- 3. Cascade-based attacks on complex networks
- 4. 【翻譯】Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks
- 5. [S&P 2019翻譯]Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks
- 6. 讀書筆記17：Adversarial Attacks on Neural Networks for Graph Data
- 7. 【生成對抗樣本】Simple Black-Box Adversarial Attacks on Deep Neural Networks
- 8. Stronger Data Poisoning Attacks Break Data Sanitization Defenses
- 9. Paper Notes: A Comprehensive Survey on Graph Neural Networks
- 10. [娜璋帶你讀論文] (02) SP2019-Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks
- 更多相關文章...
- • Docker Compose - Docker教程
- • ionic 手勢事件 - ionic 教程
- • RxJava操作符（一）Creating Observables
- • PHP Ajax 跨域問題最佳解決方案
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. 【翻譯】Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks
- 2. Poison Frogs! Targeted Clean-Label Poisoning Attacks on Neural Networks 論文閱讀、復現及思考
- 3. Cascade-based attacks on complex networks
- 4. 【翻譯】Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks
- 5. [S&P 2019翻譯]Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks
- 6. 讀書筆記17：Adversarial Attacks on Neural Networks for Graph Data
- 7. 【生成對抗樣本】Simple Black-Box Adversarial Attacks on Deep Neural Networks
- 8. Stronger Data Poisoning Attacks Break Data Sanitization Defenses
- 9. Paper Notes: A Comprehensive Survey on Graph Neural Networks
- 10. [娜璋帶你讀論文] (02) SP2019-Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks