Statistical Inference for Adversarial Training: Central Limit Theorems via Optimal Transport
Abstract
The purpose of this paper is to rigorously quantify the statistical and learning-theoretic properties of adversarial training models for classification. Equivalently, we establish the statistical properties of empirical optimal partial transport. Precisely, first we provide two types of central limit theorems (CLT): CLT centered at the expected empirical value, and CLT centered at the population one with smoothing. These results are based on the uniqueness of optimal potential for various equivalent optimal transport formulations, and the empirical process theory argument. For the binary setting, we indeed prove the uniqueness of optimal potential by leveraging the connection between optimal partial transport and the derived multi-marginal optimal transport formula. As byproducts, we also obtain the stability of a saddle point of the adversarial training model, and the sample complexity and concentration probability of the generalization error.
Related
- Spectral Analysis of Gram Matrices with Missing at Random Observations: Convergence, Central Limit Theorems, and Applications in Statistical Inference
- Brenier Meets Adversarial Training: Optimal Transport Geometry for Robust Learning
- Statistical Guarantees for Covariance Estimation with Adversarial Corruption under Constraints
- On the tradeoff between robustness and geometric accuracy in adversarial training
- Distributionally robust linear regression through the lens of adversarial training
BibTeX
@misc{jakwang2026statistical,
title = {{Statistical Inference for Adversarial Training: Central Limit Theorems via Optimal Transport}},
author = {Kim Jakwang and Kwon Dohyun},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.22240},
eprint = {2609.22240},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.22240},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Statistical Inference for Adversarial Training: Central Limit Theorems via Optimal Transport
AU - Kim Jakwang
AU - Kwon Dohyun
PY - 2026
DA - 2026/09/04
PB - arXiv
DO - 10.48550/arxiv.2609.22240
UR - https://arxiv.org/abs/2609.22240
AB - The purpose of this paper is to rigorously quantify the statistical and learning-theoretic properties of adversarial training models for classification. Equivalently, we establish the statistical properties of empirical optimal partial transport. Precisely, first we provide two types of central limit theorems (CLT): CLT centered at the expected empirical value, and CLT centered at the population one with smoothing. These results are based on the uniqueness of optimal potential for various equivalent optimal transport formulations, and the empirical process theory argument. For the binary setting, we indeed prove the uniqueness of optimal potential by leveraging the connection between optimal partial transport and the derived multi-marginal optimal transport formula. As byproducts, we also obtain the stability of a saddle point of the adversarial training model, and the sample complexity and concentration probability of the generalization error.
ER -