Minimax rate of convergence and the performance of empirical risk minimization in phase retrieval

Guillaume Lecué, Shahar Mendelson

Research output: Contribution to journalArticlepeer-review

11 Citations (Scopus)

Abstract

We study the performance of Empirical Risk Minimization in both noisy and noiseless phase retrieval problems, indexed by subsets of ℝn and relative to subgaussian sampling; that is, when the given data is yi = 〈ai; x02 + wi for a subgaussian random vector a, independent subgaussian noise w and a fixed but unknown x0 that belongs to a given T ⊂ ℝn. We show that ERM performed in T produces ẍ whose Euclidean distance to either x0 or x0 depends on the gaussian mean-width of T and on the signal-to-noise atio of the problem. The bound coincides with the one for linear regression when ǁx0ǁ2 is of the order of a constant. In addition, we obtain a sharp lower bound for the phase retrieval problem. As examples, we study the class of d-sparse vectors in ℝn and the unit ball in ∫n1.

Original languageEnglish
Article number57
Pages (from-to)1-27
Number of pages27
JournalElectronic Journal of Probability
Volume20
DOIs
Publication statusPublished - 2015
Externally publishedYes

Fingerprint

Dive into the research topics of 'Minimax rate of convergence and the performance of empirical risk minimization in phase retrieval'. Together they form a unique fingerprint.

Cite this