AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

Connecting Certified and Adversarial Training

Yuhao Mao, Mark Niklas Mueller, Marc Fischer, Martin Vechev

NeurIPS 2023 · Adversarial robustness and security › Certified and empirical robustness · approach: Certified training and ensembles · 4 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add TAPS gradient weight w_TAPS=5 to IBP loss (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 93.15 -> 93.62Improved
increase w_TAPS from 5 to 20 (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 93.62 -> 82.67Hurt
increase w_TAPS from 5 to 15 (ablation)TAPS · MNIST eps=0.3adversarial accuracy (%): 94.01 -> 95.00Improved
use w_TAPS=0 (IBP with TAPS-weighted gradients) vs IBP alone (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 93.15 -> 93.06No clear effect
multi-estimator instead of single-estimator PGD (3 ReLUs in classifier) (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 92.91 -> 93.03Improved
multi-estimator instead of single-estimator PGD (1 ReLU in classifier) (ablation)TAPS · MNIST eps=0.3natural accuracy (%): 31.47 -> 97.94Improved
increase PGD attack steps from 1 to 20 (1 restart) (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 93.36 -> 93.62Improved
increase PGD attack steps from 20 to 100 (1 restart) (ablation)TAPS · MNIST eps=0.3certified accuracy (%): 93.62 -> 93.46No clear effect
move split from 0 to 6 ReLUs in classifier (ablation)TAPS · CIFAR-10 eps=2/255certified accuracy (MN-BaB) (%): 56.14 -> 61.11Improved
move split from 0 to 6 ReLUs in classifier (ablation)TAPS · CIFAR-10 eps=2/255certified accuracy (IBP) (%): 53.54 -> 24.90Hurt
move split from 0 to 6 ReLUs in classifier (ablation)TAPS · CIFAR-10 eps=8/255certified accuracy (MN-BaB) (%): 34.60 -> 34.44No clear effect
move split from 0 to 6 ReLUs in classifier (ablation)TAPS · MNIST eps=0.3certified accuracy (IBP) (%): 93.08 -> 85.03Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy