Precise statistical analysis of classification accuracies for adversarial training

DOI10.1214/22-AOS2180MaRDI QIDQ2091832

Adel Javanmard, Mahdi Soltanolkotabi

Publication date: 2 November 2022

Published in: The Annals of Statistics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2010.11213

zbMATH Keywords

binary classificationadversarial trainingprecise high-dimensional asymptotics

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Asymptotic distribution theory in statistics (62E20) Generalized linear models (logistic models) (62J12)

Related Items (1)

Sharp global convergence guarantees for iterative nonconvex optimization with random data

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• High dimensional robust M-estimation: asymptotic variance via approximate message passing
• On the impact of predictor geometry on the performance on high-dimensional ridge-regularized generalized robust regression estimators
• On the distribution of the largest eigenvalue in principal components analysis
• The distribution of the Lasso: uniform control over sparse balls and adaptive parameter tuning
• Fundamental barriers to high-dimensional regression with convex penalties
• Surprises in high-dimensional ridgeless least squares interpolation
• The phase transition for the existence of the maximum likelihood estimate in high-dimensional logistic regression
• Optimal shrinkage of eigenvalues in the spiked covariance model
• Sparse PCA via Covariance Thresholding
• Phase transitions in semidefinite relaxations
• Precise Error Analysis of Regularized <inline-formula> <tex-math notation="LaTeX">$M$ </tex-math> </inline-formula>-Estimators in High Dimensions
• Adversarial Risk via Optimal Transport and Optimal Couplings
• The Generalization Error of Random Features Regression: Precise Asymptotics and the Double Descent Curve
• Does SLOPE outperform bridge regression?
• On the Adversarial Robustness of Robust Estimators
• Algorithmic Analysis and Statistical Estimation of SLOPE via Approximate Message Passing
• Living on the edge: phase transitions in convex programs with random data
• Reconciling modern machine-learning practice and the classical bias–variance trade-off
• A modern maximum-likelihood theory for high-dimensional logistic regression
• The LASSO Risk for Gaussian Matrices
• The Noise-Sensitivity Phase Transition in Compressed Sensing
• Information-Theoretically Optimal Compressed Sensing via Spatial Coupling and Approximate Message Passing
• Sharp Time–Data Tradeoffs for Linear Inverse Problems

This page was built for publication: Precise statistical analysis of classification accuracies for adversarial training