14
0

Compressive Recovery of Sparse Precision Matrices

Abstract

We consider the problem of learning a graph modeling the statistical relations of the dd variables from a dataset with nn samples XRn×dX \in \mathbb{R}^{n \times d}. Standard approaches amount to searching for a precision matrix Θ\Theta representative of a Gaussian graphical model that adequately explains the data. However, most maximum likelihood-based estimators usually require storing the d2d^{2} values of the empirical covariance matrix, which can become prohibitive in a high-dimensional setting. In this work, we adopt a compressive viewpoint and aim to estimate a sparse Θ\Theta from a \emph{sketch} of the data, i.e. a low-dimensional vector of size md2m \ll d^{2} carefully designed from XX using non-linear random features. Under certain assumptions on the spectrum of Θ\Theta (or its condition number), we show that it is possible to estimate it from a sketch of size m=Ω((d+2k)log(d))m=\Omega\left((d+2k)\log(d)\right) where kk is the maximal number of edges of the underlying graph. These information-theoretic guarantees are inspired by compressed sensing theory and involve restricted isometry properties and instance optimal decoders. We investigate the possibility of achieving practical recovery with an iterative algorithm based on the graphical lasso, viewed as a specific denoiser. We compare our approach and graphical lasso on synthetic datasets, demonstrating its favorable performance even when the dataset is compressed.

View on arXiv
Comments on this paper