Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > eess > arXiv:1811.04356

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Electrical Engineering and Systems Science > Signal Processing

arXiv:1811.04356 (eess)
[Submitted on 11 Nov 2018 (v1), last revised 24 Feb 2019 (this version, v2)]

Title:Bayesian Convolutional Neural Networks for Compressed Sensing Restoration

Authors:Xinjie Lan, Xin Guo, Kenneth E. Barner
View a PDF of the paper titled Bayesian Convolutional Neural Networks for Compressed Sensing Restoration, by Xinjie Lan and 2 other authors
View PDF
Abstract:Deep Neural Networks (DNNs) have aroused great attention in Compressed Sensing (CS) restoration. However, the working mechanism of DNNs is not explainable, thereby it is unclear that how to design an optimal DNNs for CS restoration. In this paper, we propose a novel statistical framework to explain DNNs, which proves that the hidden layers of DNNs are equivalent to Gibbs distributions and interprets DNNs as a Bayesian hierarchical model. The framework provides a Bayesian perspective to explain the working mechanism of DNNs, namely some hidden layers learn a prior distribution and other layers learn a likelihood distribution. Moreover, the framework provides insights into DNNs and reveals two inherent limitations of DNNs for CS restoration. In contrast to most previous works designing an end-to-end DNNs for CS restoration, we propose a novel DNNs to model a prior distribution only, which can circumvent the limitations of DNNs. Given the prior distribution generated from the DNNs, we design a Bayesian inference algorithm to realize CS restoration in the framework of Bayesian Compressed Sensing. Finally, extensive simulations validate the proposed theory of DNNs and demonstrate that the proposed algorithm outperforms the state-of-the-art CS restoration methods.
Subjects: Signal Processing (eess.SP)
Cite as: arXiv:1811.04356 [eess.SP]
  (or arXiv:1811.04356v2 [eess.SP] for this version)
  https://doi.org/10.48550/arXiv.1811.04356
arXiv-issued DOI via DataCite

Submission history

From: Xinjie Lan [view email]
[v1] Sun, 11 Nov 2018 05:31:22 UTC (2,506 KB)
[v2] Sun, 24 Feb 2019 06:01:38 UTC (1,338 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Bayesian Convolutional Neural Networks for Compressed Sensing Restoration, by Xinjie Lan and 2 other authors
  • View PDF
  • TeX Source
  • Other Formats
view license
Current browse context:
eess.SP
< prev   |   next >
new | recent | 2018-11
Change to browse by:
eess

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack