E-Mail Alert

Add your e-mail address to receive forthcoming issues of this journal:

Journal Browser

Journal Browser

Special Issue "Information Theory and Graphical Models"

A special issue of Entropy (ISSN 1099-4300). This special issue belongs to the section "Information Theory".

Deadline for manuscript submissions: closed (15 December 2017)

Special Issue Editor

Guest Editor
Prof. Dr. Teemu Roos

Department of Computer Science and Helsinki Institute for Information Technology HIIT, Exactum building, A322, PO Box 68, FI-00014 University of Helsinki, Finland
Website | E-Mail
Interests: machine learning and big data; computational statistics; probabilistic graphical models; information theory; digital humanities

Special Issue Information

Dear Colleagues,

Graphical models are an elegant framework for representing, learning, and manipulating multivariate distributions. They are used in all areas of statistics, machine learning, and artificial intelligence. Graphical models have also had a pivotal role in information theory from its early days on until the present day: One can mention examples from Markov chains and other sequence models to graph-based LDPC codes and network information theory, and beyond.

This Special Issue brings together the current trends at the intersection of information theory and graphical models with an emphasis on fundamental limits and new theoretical concepts that have potential to change the way we approach modelling problems and to bring forth new technical solutions in various applications.

We welcome contributions in mathematics, statistics, computer science and engineering, natural sciences, and philosophy, as well as novel applications of information theory and graphical models in new areas.

Prof. Dr. Teemu Roos
Guest Editor

Keywords

  • graph-based source and channel coding
  • network information theory
  • information-theoretic limits of learning in graphical models
  • minimax estimation in graphical models
  • information criteria for model selection
  • maximum entropy methods for graphical models
  • information geometry on graphs

Published Papers (1 paper)

View options order results:
result details:
Displaying articles 1-1
Export citation of selected articles as:

Research

Open AccessArticle Construction of New Fractional Repetition Codes from Relative Difference Sets with λ=1
Entropy 2017, 19(10), 563; doi:10.3390/e19100563
Received: 12 August 2017 / Revised: 19 October 2017 / Accepted: 19 October 2017 / Published: 22 October 2017
PDF Full-text (736 KB) | HTML Full-text | XML Full-text
Abstract
Fractional repetition (FR) codes are a class of distributed storage codes that replicate and distribute information data over several nodes for easy repair, as well as efficient reconstruction. In this paper, we propose three new constructions of FR codes based on relative difference
[...] Read more.
Fractional repetition (FR) codes are a class of distributed storage codes that replicate and distribute information data over several nodes for easy repair, as well as efficient reconstruction. In this paper, we propose three new constructions of FR codes based on relative difference sets (RDSs) with λ = 1 . Specifically, we propose new ( q 2 - 1 , q , q ) FR codes using cyclic RDS with parameters ( q + 1 , q - 1 , q , 1 ) constructed from q-ary m-sequences of period q 2 - 1 for a prime power q, ( p 2 , p , p ) FR codes using non-cyclic RDS with parameters ( p , p , p , 1 ) for an odd prime p or p = 4 and ( 4 l , 2 l , 2 l ) FR codes using non-cyclic RDS with parameters ( 2 l , 2 l , 2 l , 1 ) constructed from the Galois ring for a positive integer l. They are differentiated from the existing FR codes with respect to the constructable code parameters. It turns out that the proposed FR codes are (near) optimal for some parameters in terms of the FR capacity bound. Especially, ( 8 , 3 , 3 ) and ( 9 , 3 , 3 ) FR codes are optimal, that is, they meet the FR capacity bound for all k. To support various code parameters, we modify the proposed ( q 2 - 1 , q , q ) FR codes using decimation by a factor of the code length q 2 - 1 , which also gives us new good FR codes. Full article
(This article belongs to the Special Issue Information Theory and Graphical Models)
Figures

Figure 1

Back to Top