Shrinking Artificial Intelligence: Energy concerns push AI optimizations to the edge.

The article examines the energy consumption of artificial intelligence technology with a focus on the relationship between deep neural networks and Moore's Law. Energy consumption and latency of logic circuits are compared with that of memory transfers within an edge computing context along with oth...

Descripción completa

Detalles Bibliográficos
Publicado en:Communications of the ACM Vol. 65; no. 1; pp. 12 - 15
Autor principal: Edwards, Chris
Formato: Artículo
Publicado: Association for Computing Machinery Jan2022
Materias:
Acceso en línea:Ver este registro en EBSCOhost
fields @attributes:
  recordID: 1
pdfLink:
plink: https://search.ebscohost.com/login.aspx?direct=true&db=hlh&AN=154455685&site=ehost-live
header:
  @attributes:
    shortDbName: hlh
    uiTerm: 154455685
    longDbName: Humanities International Complete
    uiTag: AN
  controlInfo:
    bkinfo:
    jinfo:
      jid:
        00010782
        ACM
      jtl: Communications of the ACM
      issn: 00010782
      maglogo: N
    pubinfo:
      dt: Jan2022
      vid: 65
      iid: 1
      pid: 68
      pub: Association for Computing Machinery
    artinfo:
      ui:
        154455685
        10.1145/3495562
      ppf: 12
      ppct: 3
      formats:
      tig:
        atl: Shrinking Artificial Intelligence: Energy concerns push AI optimizations to the edge.
      aug:
        au: Edwards, Chris
      su:
        Artificial intelligence
        Energy consumption
        Moore's law
        Artificial neural networks
        Data compression
      sug:
        subj:
          Artificial intelligence
          Energy consumption
          Moore's law
          Artificial neural networks
          Data compression
      ab: The article examines the energy consumption of artificial intelligence technology with a focus on the relationship between deep neural networks and Moore's Law. Energy consumption and latency of logic circuits are compared with that of memory transfers within an edge computing context along with other strategies to reduce the energy consumption of software execution such as data compression, precision reduction, and elimination of neural redundancy.
      pubtype: Periodical
      doctype: Article
      src: R
    language: English
    refInfo:
    copyright:
      @attributes:
        flag: Y
      dt:
        @attributes:
          year: 2022
    holdings:
      @attributes:
        islocal: N