Research map / Federated learning and privacy / Differential privacy

Optimal Algorithms for Fixed-Graph Multi-Attribution Privacy

NeurIPS 2026 · Federated learning and privacy › Differential privacy · approach: Private online and user-level learning

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Studies fixed-graph differential privacy for learning from multi-attribution data, where the attribution structure is public and example contents are private, and derives fundamental limits and optimal algorithms.

Problem

Standard differential privacy assumes each training example affects only one user's privacy, but many real datasets attribute examples to multiple users; the public attribution structure must be handled while example contents stay private.

Approach

Introduces a fixed-graph differential privacy framework in which the attribution graph between examples and users is public while example contents are private, and establishes fundamental limits and optimal algorithms under this model.

Experimental setup

Builds on: differential privacy

Closest papers by meaning

• It's My Data Too: Private ML for Datasets with Multi-User Training Examples (ICML 2025 · similarity 0.90)
• Locally Optimal Private Sampling: Beyond the Global Minimax (NeurIPS 2025 · similarity 0.86)
• Differentially Private Decentralized Learning with Random Walks (ICML 2024 · similarity 0.86)
• Certification for Differentially Private Prediction in Gradient-Based Training (ICML 2025 · similarity 0.86)
• Mean Estimation in the Add-Remove Model of Differential Privacy (ICML 2024 · similarity 0.86)
• On the Price of Differential Privacy for Hierarchical Clustering (ICLR 2025 · similarity 0.86)