Can R'enyi Accountants Beat Lossless Accountants for DP-SGD?

cs.LG, cs.CR, cs.DS, cs.IT, math.IT, stat.ML

Submitted: 2026-10-12

Updated: 2026-10-12

Comments: 41 pages

Code: https://github.com/jinhobok/renyi

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Sources

• Publishing Wikipedia usage data with strong privacy guarantees
• Less Random, More Private: What is the Optimal Subsampling Scheme for DP-SGD?
• Concentrated Differential Privacy
• Privacy of the last iterate in cyclically-sampled DP-SGD on nonconvex composite losses
• JAX-Privacy: A library for differentially private machine learning
• R'enyi Differential Privacy of the Sampled Gaussian Mechanism
• Privacy Filters are Captured by Residues: A Characterization of Free Natural Filters and the Cost of Adaptivity
• Opacus: User-Friendly Differential Privacy Library in PyTorch

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks