AI Paper Atlas

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

Adversarial Training from Mean Field Perspective

Soichiro Kumano, Hiroshi Kera, Toshihiko Yamasaki

NeurIPS 2023 · Adversarial robustness and security › Certified and empirical robustness · approach: Provable robust learning guarantees · 0 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
increase network width from 256 to 512 in adversarial trainingvanilla ReLU network, L=20, eps=0.3 · MNIST training accuracyadversarial trainability: N = 256 -> 512Improved
add shortcuts (residual connections) to the networkdeep fully connected ReLU-like network · adversarial training trainabilitygradient trainabilityImproved
use adversarial training instead of standard trainingvanilla network L=10, p=inf, q=inf, eps=0.3 · MNISTweight varianceHurt
increase width N from 1k to 2kadversarial training of vanilla network · MNIST weight varianceweight variance decrease: N = 1k -> N = 2kImproved
increase depth L under adversarial trainingvanilla network · MNIST training accuracytraining accuracyHurt
use standard training instead of adversarial trainingvanilla networks of various depth and width · MNIST training accuracytraining accuracyImproved
increase network width in adversarial trainingfully trained networks · MNIST, Fashion-MNISTadversarial robustnessImproved
increase training steps T from 100 to 160 in trainability boundaryvanilla networks p=inf q=inf eps=0.3 · MNIST depth-width heat maptrainability boundary: T = 100, 160, 250Mixed
increase input/output dimension for (p,q)=(2,inf)adversarial loss upper bound · random networksupper boundNo clear effect
use finite network width instead of infiniteThm 5.1 bounds · random networks N=40,000adversarial loss vs boundMixed

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