Learning Robust and Privacy-Preserving Representations via Information Theory
Not on the map. Why?
Abstract (arXiv)
Machine learning models are vulnerable to both security attacks (e.g., adversarial examples) and privacy attacks (e.g., private attribute inference). We take the first step to mitigate both the security and privacy attacks, and maintain task utility as well. Particularly, we propose an information-theoretic framework to achieve the goals through the lens of representation learning, i.e., learning representations that are robust to both adversarial examples and attribute inference adversaries. We also derive novel theoretical results under our framework, e.g., the inherent trade-off between adversarial robustness/utility and attribute privacy, and guaranteed attribute privacy leakage against attribute inference adversaries.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 44 of its 63 references that are in the map. First posted on arXiv 15 Dec 2024.
Builds on
44 of 63 references.
References · 44
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
- 2013Intriguing properties of neural networkscited by 5,155
- 2014Deep Learning Face Attributes in the Wildcited by 3,693
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
- 2019Theoretically Principled Trade-off between Robustness and Accuracycited by 1,517
- 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
- 2018Learning deep representations by mutual information estimation and maximizationcited by 1,273
- 2016Adversarial Machine Learning at Scalecited by 936
- 2016Deep Variational Information Bottleneckcited by 930
- 2020Fast is better than free: Revisiting adversarial trainingcited by 727
- 2018MINE: Mutual Information Neural Estimationcited by 683
- 2017Provable defenses against adversarial examples via the convex outer adversarial polytopecited by 567
32 more
- 2019On Variational Bounds of Mutual Informationcited by 559
- 2018Exploiting Unintended Feature Leakage in Collaborative Learningcited by 520
- 2016f-GAN: Training Generative Neural Samplers using Variational Divergence Minimizationcited by 506
- 2018Certified Robustness to Adversarial Examples with Differential Privacycited by 468
- 2020CLUB: A Contrastive Log-ratio Upper Bound of Mutual Informationcited by 235
- 2019Improving Adversarial Robustness via Promoting Ensemble Diversitycited by 213
- 2018Formal Limitations on the Measurement of Mutual Informationcited by 189
- 2019Privacy Risks of Securing Machine Learning Models against Adversarial Examplescited by 141
- 2020MACER: Attack-free and Scalable Robust Training via Maximizing Certified Radiuscited by 114
- 2018Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flowcited by 105
- 2020Adversarial Distributional Training for Robust Deep Learningcited by 79
- 2019Metric Learning for Adversarial Robustnesscited by 77
- 2022SoK: Let the Privacy Games Begin! A Unified Treatment of Data Inference Privacy in Machine Learningcited by 51
- 2018Towards Privacy-Preserving Visual Recognition via Adversarial Training: A Pilot Studycited by 48
- 2019Mitigating Information Leakage in Image Representations: A Maximum Entropy Approachcited by 48
- 2017Adversarial Image Perturbation for Privacy Protection -- A Game Theory Perspectivecited by 38
- 2023Text-CRS: A Generalized Certified Robustness Framework against Textual Adversarial Attackscited by 34
- 2018Deep Private-Feature Extractioncited by 30
- 2018Learning Privacy Preserving Encodings through Adversarial Trainingcited by 30
- 2021Towards Defending against Adversarial Examples via Attack-Invariant Featurescited by 25
- 2021Privacy-Preserving Representation Learning on Graphs: A Mutual Information Perspectivecited by 23
- 2020Information Obfuscation of Graph Neural Networkscited by 19
- 2016Minimax Filter: Learning to Preserve Privacy from Inference Attackscited by 16
- 2022Improving Adversarial Robustness via Mutual Information Estimationcited by 13
- 2023Task-Agnostic Privacy-Preserving Representation Learning for Federated Learning Against Attribute Inference Attackscited by 13
- 2024Inf2Guard: An Information-Theoretic Framework for Learning Privacy-Preserving Representations against Inference Attackscited by 13
- 2022UniCR: Universally Approximated Certified Robustness via Randomized Smoothingcited by 10
- 2020Can we Generalize and Distribute Private Representation Learning?cited by 5
- 2023A Certified Radius-Guided Attack Framework to Image Segmentation Modelscited by 5
- 2023Certifiable Black-Box Attacks with Randomized Adversarial Examples: Breaking Defenses with Provable Confidencecited by 5
- 2022NeuGuard: Lightweight Neuron-Guided Defense against Membership Inference Attackscited by 3
- 2024Universally Harmonizing Differential Privacy Mechanisms for Federated Learning: Boosting Accuracy and Convergencecited by 1
Built on by
Citing papers per year: 2025: 3.
3 papers.