• DocumentCode
    2821698
  • Title

    Two architectures implementing feed-forward completely connected neural nets

  • Author

    Stefanelli, Renato

  • Author_Institution
    Dept. of Electron., Politecnico di Milano, Italy
  • fYear
    1991
  • fDate
    11-14 Jun 1991
  • Firstpage
    2514
  • Abstract
    Feedforward completely-connected neural nets represent the most general scheme of loopless ANNs (artificial neural networks). Two architectures are presented. The second one (unfolded) presents a higher throughput (a higher number of shorter W-SRs) but a lower regularity (two different PEs, or processing elements); the first one (folded) presents a lower number of PEs (although more complex) and a higher regularity in the output signals ordered on the same output line from n0 to nN
  • Keywords
    computer architecture; neural nets; artificial neural networks; feed-forward completely connected neural nets; folded architecture; loopless nets; regularity; throughput; unfolded architecture; Artificial neural networks; Feedforward neural networks; Feedforward systems; Feeds; Neural networks; Neurons; Pipeline processing; Signal generators; Silicon; Throughput;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Circuits and Systems, 1991., IEEE International Sympoisum on
  • Print_ISBN
    0-7803-0050-5
  • Type

    conf

  • DOI
    10.1109/ISCAS.1991.176038
  • Filename
    176038
    • Link To Document
    https://search.isc.ac/dl/search/defaultta.aspx?DTC=49&DC=2821698