Private and polynomial time algorithms for learning Gaussians and beyond

11/22/2021
by   Hassan Ashtiani, et al.
0

We present a fairly general framework for reducing (ε, δ) differentially private (DP) statistical estimation to its non-private counterpart. As the main application of this framework, we give a polynomial time and (ε,δ)-DP algorithm for learning (unrestricted) Gaussian distributions in ℝ^d. The sample complexity of our approach for learning the Gaussian up to total variation distance α is O(d^2/α^2+d^2 √(ln1/δ)/αε), matching (up to logarithmic factors) the best known information-theoretic (non-efficient) sample complexity upper bound of Aden-Ali, Ashtiani, Kamath (ALT'21). In an independent work, Kamath, Mouzakis, Singhal, Steinke, and Ullman (arXiv:2111.04609) proved a similar result using a different approach and with O(d^5/2) sample complexity dependence on d. As another application of our framework, we provide the first polynomial time (ε, δ)-DP algorithm for robust learning of (unrestricted) Gaussians.

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 1

page 2

page 3

page 4

05/01/2018

Privately Learning High-Dimensional Distributions

We design nearly optimal differentially private algorithms for learning ...
07/12/2021

Forster Decomposition and Learning Halfspaces with Noise

A Forster transform is an operation that turns a distribution into one w...
12/16/2018

Provable limitations of deep learning

As the success of deep learning reaches more grounds, one would like to ...
10/11/2021

Differentially Private Approximate Quantiles

In this work we study the problem of differentially private (DP) quantil...
02/10/2021

Agnostic Proper Learning of Halfspaces under Gaussian Marginals

We study the problem of agnostically learning halfspaces under the Gauss...
02/24/2020

Learning Structured Distributions From Untrusted Batches: Faster and Simpler

We revisit the problem of learning from untrusted batches introduced by ...
12/05/2019

On the Sample Complexity of Learning Sum-Product Networks

Sum-Product Networks (SPNs) can be regarded as a form of deep graphical ...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.