Title |
|---|
 Initialization Matters: Privacy-Utility Analysis of Overparameterized
Neural NetworksNeural Information Processing Systems (NeurIPS), 2023 |
 Gradients Look Alike: Sensitivity is Often Overestimated in DP-SGDUSENIX Security Symposium (USENIX Security), 2023 |
 A Randomized Approach for Tight Privacy AccountingNeural Information Processing Systems (NeurIPS), 2023 |
 Bounding Training Data Reconstruction in DP-SGDNeural Information Processing Systems (NeurIPS), 2023 |
 Analyzing Privacy Leakage in Machine Learning via Multiple Hypothesis
Testing: A Lesson From FanoInternational Conference on Machine Learning (ICML), 2022 |
 CANIFE: Crafting Canaries for Empirical Privacy Measurement in Federated
LearningInternational Conference on Learning Representations (ICLR), 2022 |
 Measuring Forgetting of Memorized Training ExamplesInternational Conference on Learning Representations (ICLR), 2022 |
