ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2305.04871
25
5

Gaussian process deconvolution

8 May 2023
Felipe A. Tobar
Arnaud Robert
Jorge F. Silva
ArXivPDFHTML
Abstract

Let us consider the deconvolution problem, that is, to recover a latent source x(⋅)x(\cdot)x(⋅) from the observations y=[y1,…,yN]\mathbf{y} = [y_1,\ldots,y_N]y=[y1​,…,yN​] of a convolution process y=x⋆h+ηy = x\star h + \etay=x⋆h+η, where η\etaη is an additive noise, the observations in y\mathbf{y}y might have missing parts with respect to yyy, and the filter hhh could be unknown. We propose a novel strategy to address this task when xxx is a continuous-time signal: we adopt a Gaussian process (GP) prior on the source xxx, which allows for closed-form Bayesian nonparametric deconvolution. We first analyse the direct model to establish the conditions under which the model is well defined. Then, we turn to the inverse problem, where we study i) some necessary conditions under which Bayesian deconvolution is feasible, and ii) to which extent the filter hhh can be learnt from data or approximated for the blind deconvolution case. The proposed approach, termed Gaussian process deconvolution (GPDC) is compared to other deconvolution methods conceptually, via illustrative examples, and using real-world datasets.

View on arXiv
Comments on this paper