(2020年8月29日更新)
日程 2020年9月2日(水) ~ 9月3日(木)
会場オンライン開催
主催 電子情報通信学会 情報理論とその応用サブソサイエティ
併催 電子情報通信学会 情報理論研究会
実行委員会 眞田 亜紀子(湘南工科大学,委員長), 堀井 俊佑(早稲田大学), 安永 憲司(大阪大学), 中原 悠太(早稲田大学), 瀧田 愼(兵庫県立大学)

 

開催趣旨

プログラム

本年の実施形態

参加申し込み

講演申し込み

会場について

オンライン会場

  • 参加申込された方にメールにてご案内します
  • スケジュール

    依頼講演

    以下の依頼講演を予定しています(敬称略).


    依頼講演1:9月2日(水) 14:10–15:00
    【講演者】路 姍 (Shan Lu)(岐阜大学)
    【題目】Learning-based approach for designing error-correcting codes
    【概要】Classic error-correcting code design is based on coding-theoretic principles to optimize some performance-related properties such as minimum Hamming distance, decoding threshold. The learning-based approach to designing error-correcting code aims to optimize methods to learn the parameters of the code within a certain code family. In this talk, firstly, I will introduce the basic concepts of machine learning and learning-based approaches. Then, some learning-based approaches, such as reinforcement learning and genetic algorithms will be introduced. Finally, the neural network architectures for designing error-correcting codes and corresponding training methodologies will be introduced.


    依頼講演2:9月2日(水) 16:20–17:10
    【講演者】Oliver Wilhelm Gnilke (Aalborg University, Denmark)
    【題目】Codes for Privacy and Reliability in Information Retrieval and Distributed Computation
    【概要】Linear codes have recently found applications in information theoretically secure information retrieval protocols. These schemes allow a user to download a file from a storage system consisting of a group of servers, without revealing the identity of the requested file. A very general class of schemes is based on evaluation codes. It can be shown that these schemes are asymptotically optimal. Furthermore, they are easily extended to a wide array of variants to the model, including collusion of servers, non-responsive servers, and byzantine attacks. Distributed computation is a class of related problems in which not information but the result of a computation is desired. Most commonly a user wants to distribute the task of multiplying two matrices over a set of helper nodes, without revealing the content of the matrices or the result. Polynomial codes can be used to describe schemes that use few helper nodes while preserving privacy. Similar extensions to the model as in the information retrieval case can be considered, but less is known about capacity in such scenarios.


    依頼講演3:9月3日(木) 10:00–10:50
    【講演者】Danny Dubé (Université Laval, Canada)
    【題目】Arbitrarily Low Redundancy Construction of Balanced Codes Using Realistic Resources
    【概要】We present a technique based on permutations, the well known arcade game Pacman, and limited-precision integers to encode data into balanced codewords. The redundancy that is introduced by the encoding is particularly low. The results are noticeably better than those of previous work. Still, the resources required by our technique remain modest: there is no need for costly calculations using large integers and the time and space complexity for encoding or decoding a block is linear. Although the basic technique allows one to achieve the best per-block redundancy, the per-block aspect of the technique prevents the achievement of the optimal redundancy, in a global sense. The last part of the talk presents an extension of the technique that allows one to arbitrarily approach the optimal global redundancy.


    お問い合わせ

    本ワークショップに関するお問い合わせは,実行委員会アドレス ECC-WS2020@list.waseda. (@より前のアルファベットを小文字にして末尾にjpを付加してください)宛にお願いします.


    ECCWS2012  ECCWS2013  ECCWS2014  ECCWS2015
    ECCWS2016  ECCWS2017  ECCWS2018  ECCWS2019