Near-Optimal Recovery of Linear and N-Convex Functions on Unions of Convex Sets
share
In this paper, following the line of research on "statistical inference via convex optimization" (D. Donoho, Ann. Stat. 22(1) (1994), A. Juditsky, A. Nemirovski, Ann. Stat. 37(5a) (2009), A. Goldenshluger, A. Juditsky, A. Nemirovski, Electr. J. Stat. 9(2) (2015)), we build provably near-optimal, in the minimax sense, estimates of linear forms (or more general "N-convex" functionals, the simplest example being the maximum of several fractional-linear functions) of unknown "signal" known to belong to the union of finitely many convex compact sets from indirect noisy observations of the signal. Our main assumption is that the observation scheme in question is good in the sense of A. Goldenshluger, A. Juditsky, A. Nemirovski, Electr. J. Stat. 9(2) (2015), the simplest example being the Gaussian observation scheme, where the observation is the sum of linear image of the signal and the standard Gaussian noise. The proposed estimates, same as upper bounds on their worst-case risks, stem from solutions to explicit convex optimization problems, making the estimates "computation-friendly."
READ FULL TEXT
