Buchumschlag

Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data

Fuente: arXiv

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Li, Binghui, Li, Yuanzhi
Format:	Preprint
Veröffentlicht:	2024
Schlagworte:	Machine Learning
Online-Zugang:	Acceder al recurso
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Ähnliche Einträge

On the Clean Generalization and Robust Overfitting in Adversarial Training from Two Theoretical Views: Representation Complexity and Training Dynamics
von: Li, Binghui, et al.
Veröffentlicht: (2023)

Provably learning a multi-head attention layer
von: Chen, Sitan, et al.
Veröffentlicht: (2024)

Larger Datasets Can Be Repeated More: A Theoretical Analysis of Multi-Epoch Scaling in Linear Regression
von: Yan, Tingkai, et al.
Veröffentlicht: (2025)

Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks
von: Li, Binghui, et al.
Veröffentlicht: (2024)

Certifiable Black-Box Attacks with Randomized Adversarial Examples: Breaking Defenses with Provable Confidence
von: Hong, Hanbin, et al.
Veröffentlicht: (2023)

Measure-Theoretic Anti-Causal Representation Learning
von: Behnam, Arman, et al.
Veröffentlicht: (2025)

Provable Unrestricted Adversarial Training without Compromise with Generalizability
von: Zhang, Lilin, et al.
Veröffentlicht: (2023)

Inf2Guard: An Information-Theoretic Framework for Learning Privacy-Preserving Representations against Inference Attacks
von: Noorbakhsh, Sayedeh Leila, et al.
Veröffentlicht: (2024)

You Can't Ignore Either: Unifying Structure and Feature Denoising for Robust Graph Learning
von: Yang, Tianmeng, et al.
Veröffentlicht: (2024)

Unlabeled Data Can Provably Enhance In-Context Learning of Transformers
von: Liu, Renpu, et al.
Veröffentlicht: (2026)

Adversarial Training Improves Generalization Under Distribution Shifts in Bioacoustics
von: Heinrich, René, et al.
Veröffentlicht: (2025)

Disentangling Feature Structure: A Mathematically Provable Two-Stage Training Dynamics in Transformers
von: Gong, Zixuan, et al.
Veröffentlicht: (2025)

How Does Overparameterization Affect Features?
von: Duzgun, Ahmet Cagri, et al.
Veröffentlicht: (2024)

Detecting Adversarial Data via Provable Adversarial Noise Amplification
von: Mumcu, Furkan, et al.
Veröffentlicht: (2026)

Learning Robust and Privacy-Preserving Representations via Information Theory
von: Zhang, Binghui, et al.
Veröffentlicht: (2024)

Non-Adversarial Imitation Learning Provably Free of Compounding Errors: The Role of Bellman Constraints
von: Xu, Tian, et al.
Veröffentlicht: (2026)

Provable Robustness Against a Union of $\ell_0$ Adversarial Attacks
von: Hammoudeh, Zayd, et al.
Veröffentlicht: (2023)

Feature Statistics with Uncertainty Help Adversarial Robustness
von: Wang, Ran, et al.
Veröffentlicht: (2025)

Freeze then Train: Towards Provable Representation Learning under Spurious Correlations and Feature Noise
von: Ye, Haotian, et al.
Veröffentlicht: (2022)

Physics of Language Models: Part 1, Learning Hierarchical Language Structures
von: Allen-Zhu, Zeyuan, et al.
Veröffentlicht: (2023)

Robust Satisficing Gaussian Process Bandits Under Adversarial Attacks
von: Saday, Artun, et al.
Veröffentlicht: (2025)

MEAT: Median-Ensemble Adversarial Training for Improving Robustness and Generalization
von: Hu, Zhaozhe, et al.
Veröffentlicht: (2024)

Can Mamba Learn In Context with Outliers? A Theoretical Generalization Analysis
von: Li, Hongkang, et al.
Veröffentlicht: (2025)

Provably Invincible Adversarial Attacks on Reinforcement Learning Systems: A Rate-Distortion Information-Theoretic Approach
von: Lu, Ziqing, et al.
Veröffentlicht: (2025)

Transformers Trained via Gradient Descent Can Provably Learn a Class of Teacher Models
von: Zhang, Chenyang, et al.
Veröffentlicht: (2026)

FedTilt: Towards Multi-Level Fairness-Preserving and Robust Federated Learning
von: Zhang, Binghui, et al.
Veröffentlicht: (2025)

Provable Training for Graph Contrastive Learning
von: Yu, Yue, et al.
Veröffentlicht: (2023)

TCRL: Temporal-Coupled Adversarial Training for Robust Constrained Reinforcement Learning in Worst-Case Scenarios
von: Xu, Wentao, et al.
Veröffentlicht: (2026)

Test-Time Training Provably Improves Transformers as In-context Learners
von: Gozeten, Halil Alperen, et al.
Veröffentlicht: (2025)

Data-Dependent Stability Analysis of Adversarial Training
von: Wang, Yihan, et al.
Veröffentlicht: (2024)

Muon in Associative Memory Learning: Training Dynamics and Scaling Laws
von: Li, Binghui, et al.
Veröffentlicht: (2026)

A Theoretical Analysis of Mamba's Training Dynamics: Filtering Relevant Features for Generalization in State Space Models
von: Shandirasegaran, Mugunthan, et al.
Veröffentlicht: (2026)

When and How Unlabeled Data Provably Improve In-Context Learning
von: Li, Yingcong, et al.
Veröffentlicht: (2025)

Provable Effects of Data Replay in Continual Learning: A Feature Learning Perspective
von: Ding, Meng, et al.
Veröffentlicht: (2026)

Provable Adversarial Robustness in In-Context Learning
von: Zhang, Di
Veröffentlicht: (2026)

Can Implicit Bias Imply Adversarial Robustness?
von: Min, Hancheng, et al.
Veröffentlicht: (2024)

ProFeAT: Projected Feature Adversarial Training for Self-Supervised Learning of Robust Representations
von: Addepalli, Sravanti, et al.
Veröffentlicht: (2024)

CAMP in the Odyssey: Provably Robust Reinforcement Learning with Certified Radius Maximization
von: Wang, Derui, et al.
Veröffentlicht: (2025)

Learning from Peers: Collaborative Ensemble Adversarial Training
von: Dengjin, Li, et al.
Veröffentlicht: (2025)

Provably and Practically Efficient Adversarial Imitation Learning with General Function Approximation
von: Xu, Tian, et al.
Veröffentlicht: (2024)