Timezone: »

 
Poster
SmoothMix: Training Confidence-calibrated Smoothed Classifiers for Certified Robustness
Jongheon Jeong · Sejun Park · Minkyu Kim · Heung-Chang Lee · Do-Guk Kim · Jinwoo Shin

Tue Dec 07 08:30 AM -- 10:00 AM (PST) @
in Poster Session 1 »
Randomized smoothing is currently a state-of-the-art method to construct a certifiably robust classifier from neural networks against $\ell_2$-adversarial perturbations. Under the paradigm, the robustness of a classifier is aligned with the prediction confidence, i.e., the higher confidence from a smoothed classifier implies the better robustness. This motivates us to rethink the fundamental trade-off between accuracy and robustness in terms of calibrating confidences of a smoothed classifier. In this paper, we propose a simple training scheme, coined SmoothMix, to control the robustness of smoothed classifiers via self-mixup: it trains on convex combinations of samples along the direction of adversarial perturbation for each input. The proposed procedure effectively identifies over-confident, near off-class samples as a cause of limited robustness in case of smoothed classifiers, and offers an intuitive way to adaptively set a new decision boundary between these samples for better robustness. Our experimental results demonstrate that the proposed method can significantly improve the certified $\ell_2$-robustness of smoothed classifiers compared to existing state-of-the-art robust training methods.
Keywords: Deep Learning · Machine Learning · Robustness · Adversarial Robustness and Security

Author Information

Jongheon Jeong (KAIST)
Sejun Park (KAIST)
Minkyu Kim (KAIST)
Heung-Chang Lee (kakaoenterprise)
Do-Guk Kim (Kakaoenterprise)
Jinwoo Shin (KAIST)

More from the Same Authors