arXiv research

Efficient private matrix analysis algorithms for recent variants.

Private learning of Gaussian Mixture Models without boundedness assumptions.

Private learning can perform well in high dimensions, contrary to known results.

Multi-modal data collections, such as corpora of paired images and text snippets, require analysis methods beyond single-view component and topic models. For continuous observations the current dominant approach is based on extensions of canonical correlation analysis, factorizing the variation into components shared b…

Paper optimizes private PCA for covariance estimation in statistics.

In many signal processing and machine learning applications, datasets containing private information are held at different locations, requiring the development of distributed privacy-preserving algorithms. Tensor and matrix factorizations are key components of many processing pipelines. In the distributed setting, diff…

New method improves deep CCA by modeling private components conditionally independent of common factors.

This paper achieves optimal regret bounds for locally private linear contextual bandit.

Decentralized Online Learning (online learning in decentralized networks) attracts more and more attention, since it is believed that Decentralized Online Learning can help the data providers cooperatively better solve their online problems without sharing their private data to a third party or other providers. Typical…

Differentially private hyperparameter tuning improves privacy in machine learning.

Private estimation with public data reduces sample complexity.

Principal components analysis (PCA) is a standard tool for identifying good low-dimensional approximations to data in high dimension. Many data sets of interest contain private or sensitive information about individuals. Algorithms which operate on such data should be sensitive to the privacy risks in publishing their …

The paper proposes differentially private sliced inverse regression algorithms for high-dimensional data.

This work explains how maximizing latent correlations across multiple data views helps in identifying shared and private components.

A framework for private causal effect estimation without structural assumptions.

Learning the parameters of Gaussian mixture models is a fundamental and widely studied problem with numerous applications. In this work, we give new algorithms for learning the parameters of a high-dimensional, well separated, Gaussian mixture model subject to the strong constraint of differential privacy. In particula…

This paper sharpens privacy guarantees for high-dimensional PCA under differential privacy.

Blind source separation algorithms such as independent component analysis (ICA) are widely used in the analysis of neuroimaging data. In order to leverage larger sample sizes, different data holders/sites may wish to collaboratively learn feature representations. However, such datasets are often privacy-sensitive, prec…

Google Trends data improves economic forecasts of private consumption.

We propose a new input perturbation mechanism for publishing a covariance matrix to achieve $(ε,0)$-differential privacy. Our mechanism uses a Wishart distribution to generate matrix noise. In particular, We apply this mechanism to principal component analysis. Our mechanism is able to keep the positive semi-definitene…

Paper introduces efficient top-k selection with differential privacy.

New private mean estimation method works well for anisotropic data.

Privacy-preserving reinforcement learning from human feedback using decoupled reward modeling.

DP-PCA improves privacy in PCA computations with optimal statistical error.

Improved sample efficiency for private learning of Gaussian mixtures.

Paper bridges statistical inference for DP-SGD, a privacy-preserving machine learning method.

Predicting the exit (e.g. bankrupt, acquisition, etc.) of privately held companies is a current and relevant problem for investment firms. The difficulty of the problem stems from the lack of reliable, quantitative and publicly available data. In this paper, we contribute to this endeavour by constructing an exit predi…

DOPPLER optimizes DP training with low-pass filtering, improving model accuracy.

Improved private geometric median estimation with nearly-linear time complexity.

Polynomial-time algorithm estimates mean with bounded covariance using differential privacy.

Privacy is a major issue in learning from distributed data. Recently the cryptographic literature has provided several tools for this task. However, these tools either reduce the quality/accuracy of the learning algorithm---e.g., by adding noise---or they incur a high performance penalty and/or involve trusting externa…

This paper studies trade-offs in private prediction methods.

Private method measures nonlinear correlations between data hosted across two entities.

Differentially private method for synthetic data generation from vertically partitioned data.

New method identifies shared components from unpaired multimodal mixtures.

New algorithms for privately learning decision lists and halfspaces.

Private PGB boosts synthetic data quality using GANs and privacy techniques.

Study proposes a tax-based system to share disaster risk among regions.

Near-optimal private tests for simple and MLR hypotheses developed under Gaussian differential privacy.

PEARL uses AI to replicate private equity performance with liquid assets.

New mechanism for pure differential privacy on functional summaries using Laplace-like process.

We present a provably optimal differentially private algorithm for the stochastic multi-arm bandit problem, as opposed to the private analogue of the UCB-algorithm [Mishra and Thakurta, 2015; Tossou and Dimitrakakis, 2016] which doesn't meet the recently discovered lower-bound of $Ω\left(\frac{K\log(T)}ε \right)$ [Shar…

Public pretraining improves private model training even in extreme distribution shift scenarios.

We empirically test the effects of unanticipated fiscal policy shocks on the growth rate and the cyclical component of real private output and reveal different types of asymmetries in fiscal policy implementation. The data used are quarterly U.S. observati ons over the period 1967:1 to 2011:4. In doing so, we use both …

Develops Merton's model for private companies using DDM.

Transform non-private e-values into differentially private ones.

Algorithm selects public datasets for private machine learning.

Optimizes private statistics with noisy methods.