Skip to main content

Roundtable: Machine learning for embedded systems

Authors

X. Sharon Hu, Rolf Ernst, Petru Eles, Gernot Heiser, Kurt Keutzer, Daehyun Kim and Tetsuya Tohdo

DATA61

Denso

Linköping University

University of California at Berkeley

TU Braunschweig

Samsung

University of Notre Dame

UNSW Sydney

Abstract

A previous IEEE Design&Test “Roundtable” already discussed the aspect of machine learning (ML) test and verification, but the impact of ML is wider, including hardware, software, and communication architectures and design, as well as behavioral guarantees, just to name a few important fields. ML has also started to develop a strong impact on important embedded systems design and applications. While the initial success raises high expectations for the reinvention of engineering, a discussion is overdue on where this development will eventually lead us in research and engineering. A highly attended plenary panel at Embedded Systems Week (ESWEEK) 2017 in Seoul, South Korea, with the provocative title “Machine Learning for Embedded Systems: Hype or Lasting Impact?” spurred a lively and controversial discussion that is continued in this roundtable. It is moderated by the panel organizers and moderators X. Sharon Hu, University of Notre Dame, and Rolf Ernst, Technische Universität Brunswick. Panelists include Petru Eles, Linköping University; Gernot Heiser, University of New South Wales Sydney; Kurt Keutzer, University of California at Berkeley; Daehyun Kim, Samsung; and Tetsuya Tohdo, DENSO CORP.

BibTeX Entry

  @article{Hu_EEHKKT_18,
    publisher        = {IEEE Computer Society},
    issue            = {6},
    journal          = {IEEE Design \& Test Magazine},
    author           = {Hu, X. Sharon and Ernst, Rolf and Eles, Petru and Heiser, Gernot and Keutzer, Kurt and Kim, Daehyun
                        and Tohdo, Tetsuya},
    month            = nov,
    volume           = {35},
    year             = {2018},
    date             = {2018-11-30},
    title            = {Roundtable: Machine Learning for Embedded Systems: Hype or Lasting Impact?},
    pages            = {86-93}
  }

Download