Adversarial Covariate Shift

Adversarial covariate shift focuses on improving the robustness of machine learning models against changes in the data distribution between training and deployment, particularly those changes introduced by adversarial attacks. Current research emphasizes developing algorithms and training techniques, such as adversarial training with frequency-based data augmentation and robust optimization methods, to mitigate the negative impact of these shifts. This work is crucial for building reliable and trustworthy machine learning systems, improving their performance and fairness across diverse and potentially manipulated datasets in real-world applications.

Papers

November 11, 2024

Learning a Single Neuron Robustly to Distributional Shifts and Adversarial Label Noise
Shuyao Li, Sushrut Karmalkar, Ilias Diakonikolas, Jelena Diakonikolas
Distribution Shift Primal Dual Worst Case Adversarial Noise Real Neuron Adversarial Covariate Shift

October 16, 2024

DAT: Improving Adversarial Robustness via Generative Amplitude Mix-up in Frequency Domain
Fengpeng Li, Kemou Li, Haiwei Wu, Jinyu Tian, Jiantao Zhou
Adversarial Attack Adversarial Example Adversarial Robustness Adversarial Perturbation Frequency Domain Dual Adversarial Adversarial Covariate Shift

January 22, 2024

Mitigating Covariate Shift in Misspecified Regression with Applications to Reinforcement Learning
Philip Amortila, Tongyi Cao, Akshay Krishnamurthy
Reinforcement Learning Financial Application Distribution Shift Covariate Shift Model Misspecification Adversarial Covariate Shift

May 15, 2023

Exploiting Frequency Spectrum of Adversarial Images for General Robustness
Chun Yang Tan, Kazuhiko Kawamoto, Hiroshi Kera
Adversarial Perturbation Adversarial Image General Robustness Adversarial Covariate Shift

February 23, 2023

Domain Generalisation via Domain Adaptation: An Adversarial Fourier Amplitude Approach
Minyoung Kim, Da Li, Timothy Hospedales
Domain Adaptation Domain Generalization Phase Contrast Adversarial Covariate Shift

December 5, 2022

Blessings and Curses of Covariate Shifts: Adversarial Learning Dynamics, Directional Convergence, and Equilibria
Tengyuan Liang
Adversarial Attack Adversarial Perturbation Covariate Shift Reversal Curse Efficient Equilibrium Adversarial Covariate Shift Export Import Exchange Rate Convergence

October 12, 2022

Fairness via Adversarial Attribute Neighbourhood Robust Learning
Qi Qi, Shervin Ardeshir, Yi Xu, Tianbao Yang
Procedural Fairness Fair Representation Protected Attribute Adversarial Covariate Shift Fairness Benchmark Datasets

June 1, 2022

RoCourseNet: Distributionally Robust Training of a Prediction Aware Recourse Model
Hangzhi Guo, Feiran Jia, Jinghui Chen, Anna Squicciarini, Amulya Yadav
Counterfactual Explanation Distributionally Robust Robust Recourse Adversarial Covariate Shift

April 14, 2022

Clothes-Changing Person Re-identification with RGB Modality Only
Xinqian Gu, Hong Chang, Bingpeng Ma, Shutao Bai, Shiguang Shan, Xilin Chen
Person Re Identification Video Based Person Re Identification Adversarial Covariate Shift RGB Modality

March 14, 2022

Adversarial amplitude swap towards robust image classifiers
Chun Yang Tan, Kazuhiko Kawamoto, Hiroshi Kera
Native Robustness Image Classification Adversarial Perturbation Adversarial Image Robust Overfitting General Robustness Adversarial Covariate Shift

January 28, 2022

Certifying Model Accuracy under Distribution Shifts
Aounon Kumar, Alexander Levine, Tom Goldstein, Soheil Feizi
Native Robustness Adversarial Perturbation Distribution Shift Robustness Guarantee Adversarial Covariate Shift

Adversarial Covariate Shift

Papers

Learning a Single Neuron Robustly to Distributional Shifts and Adversarial Label Noise

DAT: Improving Adversarial Robustness via Generative Amplitude Mix-up in Frequency Domain

Mitigating Covariate Shift in Misspecified Regression with Applications to Reinforcement Learning

Exploiting Frequency Spectrum of Adversarial Images for General Robustness

Domain Generalisation via Domain Adaptation: An Adversarial Fourier Amplitude Approach

Blessings and Curses of Covariate Shifts: Adversarial Learning Dynamics, Directional Convergence, and Equilibria

Fairness via Adversarial Attribute Neighbourhood Robust Learning

RoCourseNet: Distributionally Robust Training of a Prediction Aware Recourse Model

Clothes-Changing Person Re-identification with RGB Modality Only

Adversarial amplitude swap towards robust image classifiers

Certifying Model Accuracy under Distribution Shifts