• DocumentCode
    1808750
  • Title

    Implementing neural networks into modern technology

  • Author

    Johnson, M.

  • Author_Institution
    George Washington Univ., Washington, DC, USA
  • Volume
    2
  • fYear
    1999
  • fDate
    36342
  • Firstpage
    1028
  • Abstract
    The practical implementation of signal separation methods such as artificial neural networks and independent component analysis (ICA) bring several problems to light. These problems include time independence, stationarity, linearity, prior knowledge of sources, and hardware implementation. A direct emphasis is placed on implementing these networks on vocal signals, more specifically, voice-activated systems (VAS). What aspects and limitations of ICA and BSS (blind source separation) must be considered are given; the significance of preprocessing or prewhitening is considered; and learning algorithms are given. Also, a discussion is given on hardware enhancement, which may solve some of the problems related to ICA
  • Keywords
    learning (artificial intelligence); matrix algebra; neural nets; speech recognition; speech-based user interfaces; voice equipment; blind source separation; independent component analysis; learning algorithms; linearity; modern technology; preprocessing; prewhitening; signal separation methods; stationarity; time independence; vocal signals; voice-activated systems; Biomedical signal processing; Hardware; Independent component analysis; Military standards; Modems; Neural networks; Security; Signal processing algorithms; Source separation; Speech recognition;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Neural Networks, 1999. IJCNN '99. International Joint Conference on
  • Conference_Location
    Washington, DC
  • ISSN
    1098-7576
  • Print_ISBN
    0-7803-5529-6
  • Type

    conf

  • DOI
    10.1109/IJCNN.1999.831096
  • Filename
    831096