AI Paper Atlas

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

On the Global Optima of Kernelized Adversarial Representation Learning

Bashir Sadeghi, Runyi Yu, Vishnu Boddeti

ICCV 2019 · Adversarial robustness and security › Certified and empirical robustness · approach: Provable robust learning guarantees · 22 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
use closed-form Spectral-ARL (Linear-SARL) instead of DNN ML-ARLARL encoder · Adulttarget accuracy, income (%): 84.4 -> 84.1Hurt
use Linear-SARL instead of MaxEnt-ARLARL encoder · Germantarget accuracy, credit (%): 72.5 -> 76.3Improved
use Linear-SARL instead of ML-ARLARL encoder · Germantarget accuracy, credit (%): 74.4 -> 76.3Improved
use Kernel-SARL instead of Linear-SARL (ablation)SARL · Adult and Germantarget and sensitive accuracy (%): 84.1 -> 84.1No clear effect
use Linear-SARL instead of MaxEnt-ARLARL encoder · Extended Yale Badversary accuracy, illumination (%): 40 -> 21Improved
use Kernel-SARL instead of MaxEnt-ARLARL encoder · Extended Yale Btarget accuracy, identity (%): 89 -> 86Hurt
use Kernel-SARL instead of Linear-SARL (ablation)SARL · Extended Yale Btarget accuracy, identity (%): 81 -> 86Improved
exchange training and test sets (ablation)Linear-SARL · Extended Yale Btarget accuracy, identity (%): 81 -> 86Improved
use kernel (non-linear) encoder instead of linear encoder (ablation)SARL · Gaussian mixtureadversary accuracy (%): 50%Improved
use kernel encoder instead of linear encoder (ablation)SARL · CIFAR-100target vs adversary trade-offImproved
use closed-form SARL instead of simultaneous SGDARL encoder · CIFAR-100training stabilityImproved

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