Research map / Federated learning and privacy / Differential privacy

Optimal Rates for Adaptive Private $k$-PCA

NeurIPS 2026 · Federated learning and privacy › Differential privacy · approach: Local-DP and private estimation

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 differentially private stochastic k-PCA with per-sample privacy and adaptive rates, building on k-DP-PCA of Düngler & Sanyal [2025] to achieve optimal rates.

Problem

Differentially private stochastic PCA: given n i.i.d. random matrices with common expectation Sigma, find a k-dimensional subspace capturing leading variance directions while preserving per-sample DP.

Approach

Improves on k-DP-PCA by achieving optimal (adaptive) rates for private k-PCA; the abstract is truncated so the precise technique is not stated.

Limitations

• Full text unavailable; only abstract used

Experimental setup

Baselines: k-DP-PCA

Builds on: k-DP-PCA

Closest papers by meaning

• An Iterative Algorithm for Differentially Private $k$-PCA with Adaptive Noise (NeurIPS 2025 · similarity 0.97)
• On the Computational Complexity of Private High-dimensional Model Selection (NeurIPS 2024 · similarity 0.87)
• Public-data Assisted Private Stochastic Optimization: Power and Limitations (NeurIPS 2024 · similarity 0.86)
• On Inherent Privacy of Posterior Sampling: A Unified R\'enyi-Divergence Framework (NeurIPS 2026 · similarity 0.86)
• Sampling-Free Privacy Accounting for Matrix Mechanisms under Random Allocation (NeurIPS 2026 · similarity 0.86)
• Nearly-Linear Time Private Hypothesis Selection with the Optimal Approximation Factor (NeurIPS 2025 · similarity 0.86)