Optimal SQ Lower Bounds for Robustly Learning Discrete Product Distributions and Ising Models

06/09/2022
by   Ilias Diakonikolas, et al.
4

We establish optimal Statistical Query (SQ) lower bounds for robustly learning certain families of discrete high-dimensional distributions. In particular, we show that no efficient SQ algorithm with access to an ϵ-corrupted binary product distribution can learn its mean within ℓ_2-error o(ϵ√(log(1/ϵ))). Similarly, we show that no efficient SQ algorithm with access to an ϵ-corrupted ferromagnetic high-temperature Ising model can learn the model to total variation distance o(ϵlog(1/ϵ)). Our SQ lower bounds match the error guarantees of known algorithms for these problems, providing evidence that current upper bounds for these tasks are best possible. At the technical level, we develop a generic SQ lower bound for discrete high-dimensional distributions starting from low dimensional moment matching constructions that we believe will find other applications. Additionally, we introduce new ideas to analyze these moment-matching constructions for discrete univariate distributions.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/12/2022

Lower Bounds for the Total Variation Distance Given Means and Variances of Distributions

For arbitrary two probability measures on real d-space with given means ...
research
03/17/2018

Learning Mixtures of Product Distributions via Higher Multilinear Moments

Learning mixtures of k binary product distributions is a central problem...
research
07/31/2023

New Lower Bounds for Testing Monotonicity and Log Concavity of Distributions

We develop a new technique for proving distribution testing lower bounds...
research
06/19/2018

Adaptive Bayesian Estimation of Mixed Discrete-Continuous Distributions under Smoothness and Sparsity

We consider nonparametric estimation of a mixed discrete-continuous dist...
research
10/04/2018

On the Inner Product Predicate and a Generalization of Matching Vector Families

Motivated by cryptographic applications such as predicate encryption, we...
research
05/08/2019

Bounding distributional errors via density ratios

We present some new and explicit error bounds for the approximation of d...
research
04/01/2015

Learning in the Presence of Corruption

In supervised learning one wishes to identify a pattern present in a joi...

Please sign up or login with your details

Forgot password? Click here to reset