Misleading Authorship Attribution of Source Code using Adversarial Learning
一、Abstract 文章提出了一種針對源代碼做者身份的攻擊。因爲歸因方法依賴於機器學習，經過執行一系列保留語義的代碼轉換，會致使錯誤的歸因。攻擊由蒙特卡洛樹搜索引導，這樣可以在源代碼的離散域中進行操做。對204位程序員的源代碼進行評估，證實了攻擊方法對歸因方法產生了重大影響，在遭受攻擊時，其準確性從88％降低至1％。此外，文章也證實了該攻擊能夠高精度地模仿開發人員的編碼風格，從而致使錯誤的歸因。
相關文章
- 1. Large-Scale and Language-Oblivious Code Authorship Identification
- 2. Using Machine Learning Methods to Establish Program Authorship 閱讀筆記
- 3. Learning notes of developing Android using UDE
- 4. Analysis of Hello2 source code
- 5. Improving Lesion Segmentation for Diabetic Retinopathy using Adversarial Learning
- 6. 【論文筆記】 Reinforcement-Learning-Guided Source Code Summarization using Hierarchical Attention
- 7. Adversarial Sample misleading the Model(生成對抗樣本迷惑模型)
- 8. 15.Machine-Learning Supported Vulnerability Detection in Source Code
- 9. Machine Learning with Membership Privacy using Adversarial Regularization
- 10. build MYSQL source code Using visual studio 2017
- 更多相關文章...
- • RSS
元素 - RSS 教程 - • XSLT
- • Java Agent入門實戰（一）-Instrumentation介紹與使用
- • 爲了進字節跳動，我精選了29道Java經典算法題，帶詳細講解
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. Large-Scale and Language-Oblivious Code Authorship Identification
- 2. Using Machine Learning Methods to Establish Program Authorship 閱讀筆記
- 3. Learning notes of developing Android using UDE
- 4. Analysis of Hello2 source code
- 5. Improving Lesion Segmentation for Diabetic Retinopathy using Adversarial Learning
- 6. 【論文筆記】 Reinforcement-Learning-Guided Source Code Summarization using Hierarchical Attention
- 7. Adversarial Sample misleading the Model(生成對抗樣本迷惑模型)
- 8. 15.Machine-Learning Supported Vulnerability Detection in Source Code
- 9. Machine Learning with Membership Privacy using Adversarial Regularization
- 10. build MYSQL source code Using visual studio 2017