Deep k-NN Defense Against Clean-label Data Poisoning Attacks
Deep k-NN Defense Against Clean-label Data Poisoning Attacks abstract。
原理
基于 k-NN 的异常样本点检测，检验当前样本的 k 个邻近样本，如果标签不一致则标记为异常样本，进行过滤。
实现的防御方法
- L2-Norm Outlier Defense
- One-Class SVM Defense
- Random Point Eviction Defense
- Adversarial Training Defense
All articles in this blog are licensed under CC BY-NC-SA 4.0 unless stating additionally.