Max-Mahalanobis Training

[P2333]

Name: {Rethinking Softmax Cross-Entropy Loss for Adversarial Robustness }

Authors: {Tianyu Pang, Kun Xu, Yinpeng Dong, Chao Du, Ning Chen, Jun Zhu}

Paper: {https://arxiv.org/pdf/1905.10626.pdf}

Code: { https://github.com/P2333/Max-Mahalanobis-Training }

Venue: {venue if published}

Does the code implement the robust-ml API and include pre-trained models: {no}

Dataset: {cifar10}

Threat model: {ℓ∞(ϵ=8/255)}

Natural accuracy: {93.6%}

Claims: {46.0%}