Privately Learning Subspaces
Explore this paper's citation graph
Summary
Differentially private algorithms that take input data sampled from a low-dimensional linear subspace and output that subspace (or an approximation to it) and can serve as a pre-processing step for other procedures are presented.
- Type
- preprint
- Published
- 2021-05-28
- Cited by
- 23
- References
- 49
- Access
- Open access
- OpenAlex
- https://openalex.org/W3172436493
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:235265873
Keywords
Linear subspace, Subspace topology, Curse of dimensionality, Dimension (graph theory), Gradient descent
References
- Between Pure and Approximate Differential Privacy
- How To Break Anonymity of the Netflix Prize Dataset
- Private Multiplicative Weights Beyond Linear Queries
- Beating randomized response on incoherent matrices
- A Multiplicative Weights Mechanism for Privacy-Preserving Data Analysis
- Analyze gauss: optimal bounds for privacy-preserving principal component analysis
- Practical privacy: the SuLQ framework
- Fingerprinting codes and the price of approximate differential privacy
- Differentially private recommender systems: building privacy into the net
- The Johnson-Lindenstrauss Transform Itself Preserves Differential Privacy
- Pseudorandom Bits for Polynomials
- A learning theory approach to noninteractive database privacy
- Beyond worst-case analysis in private singular vector computation
- On the geometry of differential privacy
- Smooth sensitivity and sampling in private data analysis
- Near-optimal Differentially Private Principal Components
- The Noisy Power Method: A Meta Algorithm with Applications
- Robust Traceability from Trace Amounts
- Simultaneous Private Learning of Multiple Concepts
- Rate-Optimal Perturbation Bounds for Singular Subspaces with Applications to High-Dimensional Statistics
Cited by
- A Framework for Private Matrix Analysis in Sliding Window Model
- Differentially Private Covariance Revisited
- Privately Estimating a Gaussian: Efficient, Robust and Optimal
- Fast, Sample-Efficient, Affine-Invariant Private Mean and Covariance Estimation for Subgaussian Distributions
- Privately Estimating a Gaussian: Efficient, Robust, and Optimal
- PLAN: Variance-Aware Private Mean Estimation
- Smooth Lower Bounds for Differentially Private Algorithms via Padding-and-Permuting Fingerprinting Codes
- On Differentially Private Subspace Estimation in a Distribution-Free Setting
- Efficiently Computing Similarities to Private Datasets
- Does SGD really happen in tiny subspaces?
- Fedpower: privacy-preserving distributed eigenspace estimation
- Dimension-free Private Mean Estimation for Anisotropic Distributions
- On Differentially Private Linear Algebra
- Towards Reliable and Generalizable Differentially Private Machine Learning (Extended Version)
- An Iterative Algorithm for Differentially Private k-PCA with Adaptive Noise
- Homomorphic Matrix Completion
- A Private and Computationally-Efficient Estimator for Unbounded Gaussians
- (Nearly) Dimension Independent Private ERM with AdaGrad Ratesvia Publicly Estimated Subspaces
- Private and polynomial time algorithms for learning Gaussians and beyond
- FriendlyCore: Practical Differentially Private Aggregation
Related papers
- Approximate Subspace-Sparse Recovery in the Presence of Corruptions via ℓ_1-Minimization
- Fast, Robust and Non-convex Subspace Recovery
- Fast Robust Subspace Tracking via PCA in Sparse Data-Dependent Noise
- Noisy Sparse Subspace Clustering
- Subspace Detection from Structured Union of Subspaces via Linear Sampling