AI Paper Atlas

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

Defending Against Adversarial Attacks via Neural Dynamic System

Xiyuan Li, Xin Zou, Weiwei Liu

NeurIPS 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Provable robust learning guarantees · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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 the asymptotically stable neural ODE denoiser in front of the classifier (ablation)ASODE · MNIST PGDaccuracy (%): 0.36 -> 46.85Improved
add the asymptotically stable neural ODE denoiser in front of the classifier (ablation)ASODE · MNIST FGSMaccuracy (%): 9.87 -> 65.13Improved
add the asymptotically stable neural ODE denoiser in front of the classifier (ablation)ASODE · CIFAR-10 PGDaccuracy (%): 3.18 -> 57.35Improved
add the asymptotically stable neural ODE denoiser in front of the classifier (ablation)ASODE · CIFAR-10 FGSMaccuracy (%): 47.54 -> 69.94Improved
use ASODE (nonautonomous, asymptotic stability on inputs) instead of SODEFASODE · MNIST PGDaccuracy (%): 45.25 -> 46.85Improved
use ASODE instead of SODEFASODE · CIFAR-10 PGDaccuracy (%): 55.59 -> 57.35Improved
use ASODE instead of SODEFASODE · MNIST FGSMaccuracy (%): 63.36 -> 65.13Improved
use ASODE instead of SODEFASODE · CIFAR-10 FGSMaccuracy (%): 68.05 -> 69.94Improved
use ASODE instead of TisODE-NetASODE · MNIST PGDaccuracy (%): 1.78 -> 46.85Improved
use ASODE instead of TisODE-NetASODE · CIFAR-10 PGDaccuracy (%): 3.82 -> 57.35Improved
use ASODE instead of SODEF on clean inputsASODE · MNIST cleanaccuracy (%): 99.44 -> 99.44No clear effect
use ASODE instead of SODEF on clean inputsASODE · CIFAR-10 cleanaccuracy (%): 95.00 -> 95.16No clear effect

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