论文信息 - Training Ensembles to Detect Adversarial Examples

Training Ensembles to Detect Adversarial Examples

We propose a new ensemble method for detecting and classifying adversarial examples generated by state-of-the-art attacks, including DeepFool and C&W. Our method works by training the members of an ensemble to have low classification error on random benign examples while simultaneously minimizing agreement on examples outside the training distribution. We evaluate on both MNIST and CIFAR-10, against oblivious and both white- and black-box adversaries.

Razvan C. Bunescu | Gordon Stewart | Alexander Bagnall | Gordon Stewart | Alexander Bagnall

[1] Xin Yao,et al. Ensemble learning via negative correlation , 1999, Neural Networks.

[2] Dawn Xiaodong Song,et al. Adversarial Example Defenses: Ensembles of Weak Defenses are not Strong , 2017, ArXiv.

[3] Samy Bengio,et al. Adversarial examples in the physical world , 2016, ICLR.

[4] David Wagner,et al. Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methods , 2017, AISec@CCS.

[5] Seyed-Mohsen Moosavi-Dezfooli,et al. DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[6] Lior Rokach,et al. Taxonomy for characterizing ensemble methods in classification tasks: A review and annotated bibliography , 2009, Comput. Stat. Data Anal..

[7] David A. Wagner,et al. Towards Evaluating the Robustness of Neural Networks , 2016, 2017 IEEE Symposium on Security and Privacy (SP).

[8] Jonathon Shlens,et al. Explaining and Harnessing Adversarial Examples , 2014, ICLR.

[9] Joan Bruna,et al. Intriguing properties of neural networks , 2013, ICLR.