6
8

Denoising modulo samples: k-NN regression and tightness of SDP relaxation

Abstract

Many modern applications involve the acquisition of noisy modulo samples of a function ff, with the goal being to recover estimates of the original samples of ff. For a Lipschitz function f:[0,1]dRf:[0,1]^d \to \mathbb{R}, suppose we are given the samples yi=(f(xi)+ηi)mod1;i=1,,ny_i = (f(x_i) + \eta_i)\bmod 1; \quad i=1,\dots,n where ηi\eta_i denotes noise. Assuming ηi\eta_i are zero-mean i.i.d Gaussian's, and xix_i's form a uniform grid, we derive a two-stage algorithm that recovers estimates of the samples f(xi)f(x_i) with a uniform error rate O((lognn)1d+2)O((\frac{\log n}{n})^{\frac{1}{d+2}}) holding with high probability. The first stage involves embedding the points on the unit complex circle, and obtaining denoised estimates of f(xi)mod1f(x_i)\bmod 1 via a kkNN (nearest neighbor) estimator. The second stage involves a sequential unwrapping procedure which unwraps the denoised mod 11 estimates from the first stage. The estimates of the samples f(xi)f(x_i) can be subsequently utilized to construct an estimate of the function ff, with the aforementioned uniform error rate. Recently, Cucuringu and Tyagi proposed an alternative way of denoising modulo 11 data which works with their representation on the unit complex circle. They formulated a smoothness regularized least squares problem on the product manifold of unit circles, where the smoothness is measured with respect to the Laplacian of a proximity graph GG involving the xix_i's. This is a nonconvex quadratically constrained quadratic program (QCQP) hence they proposed solving its semidefinite program (SDP) based relaxation. We derive sufficient conditions under which the SDP is a tight relaxation of the QCQP. Hence under these conditions, the global solution of QCQP can be obtained in polynomial time.

View on arXiv
Comments on this paper