An improved novelty criterion for resource allocating networks

Alan McLachlan

    Research output: Chapter in Book/Report/Conference proceedingChapter

    Abstract

    Online model order complexity estimation remains one of the key problems in neural network research. The problem is further exacerbated in situations where the underlying system generator is non-stationary. In this paper, we introduce a novelty criterion for resource allocating networks (RANs) which is capable of being applied to both stationary and slowly varying non-stationary problems. The deficiencies of existing novelty criteria are discussed and the relative performances are demonstrated on two real-world problems : electricity load forecasting and exchange rate prediction.
    Original languageEnglish
    Title of host publicationFifth International Conference on Artificial Neural Networks
    PublisherIEEE
    Pages48-52
    Number of pages5
    Volume440
    ISBN (Print)0852966903
    Publication statusPublished - 7 Jul 1997
    EventFifth International Conference on Artificial Neural Networks -
    Duration: 7 Jul 19977 Jul 1997

    Publication series

    NameConference publication
    PublisherInstitute of Electrical and Electronics Engineers (IEEE)
    Volume440

    Conference

    ConferenceFifth International Conference on Artificial Neural Networks
    Period7/07/977/07/97

    Fingerprint

    Electricity
    Neural networks

    Bibliographical note

    ©1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

    Keywords

    • feedforward neural nets
    • electricity load forecasting
    • exchange rate prediction
    • extended Kalman filter training
    • algorithm
    • network growth
    • network growth prescription
    • nonstationary real-world problems
    • novelty criterion
    • radial basis function network resource allocation
    • signal processing theory
    • slowly varying nonstationary environment

    Cite this

    McLachlan, A. (1997). An improved novelty criterion for resource allocating networks. In Fifth International Conference on Artificial Neural Networks (Vol. 440, pp. 48-52). (Conference publication; Vol. 440). IEEE.
    McLachlan, Alan. / An improved novelty criterion for resource allocating networks. Fifth International Conference on Artificial Neural Networks. Vol. 440 IEEE, 1997. pp. 48-52 (Conference publication).
    @inbook{8b274e5dca6641e884e1d3e179342584,
    title = "An improved novelty criterion for resource allocating networks",
    abstract = "Online model order complexity estimation remains one of the key problems in neural network research. The problem is further exacerbated in situations where the underlying system generator is non-stationary. In this paper, we introduce a novelty criterion for resource allocating networks (RANs) which is capable of being applied to both stationary and slowly varying non-stationary problems. The deficiencies of existing novelty criteria are discussed and the relative performances are demonstrated on two real-world problems : electricity load forecasting and exchange rate prediction.",
    keywords = "feedforward neural nets, electricity load forecasting, exchange rate prediction, extended Kalman filter training, algorithm, network growth, network growth prescription, nonstationary real-world problems, novelty criterion, radial basis function network resource allocation, signal processing theory, slowly varying nonstationary environment",
    author = "Alan McLachlan",
    note = "{\circledC}1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.",
    year = "1997",
    month = "7",
    day = "7",
    language = "English",
    isbn = "0852966903",
    volume = "440",
    series = "Conference publication",
    publisher = "IEEE",
    pages = "48--52",
    booktitle = "Fifth International Conference on Artificial Neural Networks",
    address = "United States",

    }

    McLachlan, A 1997, An improved novelty criterion for resource allocating networks. in Fifth International Conference on Artificial Neural Networks. vol. 440, Conference publication, vol. 440, IEEE, pp. 48-52, Fifth International Conference on Artificial Neural Networks, 7/07/97.

    An improved novelty criterion for resource allocating networks. / McLachlan, Alan.

    Fifth International Conference on Artificial Neural Networks. Vol. 440 IEEE, 1997. p. 48-52 (Conference publication; Vol. 440).

    Research output: Chapter in Book/Report/Conference proceedingChapter

    TY - CHAP

    T1 - An improved novelty criterion for resource allocating networks

    AU - McLachlan, Alan

    N1 - ©1997 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

    PY - 1997/7/7

    Y1 - 1997/7/7

    N2 - Online model order complexity estimation remains one of the key problems in neural network research. The problem is further exacerbated in situations where the underlying system generator is non-stationary. In this paper, we introduce a novelty criterion for resource allocating networks (RANs) which is capable of being applied to both stationary and slowly varying non-stationary problems. The deficiencies of existing novelty criteria are discussed and the relative performances are demonstrated on two real-world problems : electricity load forecasting and exchange rate prediction.

    AB - Online model order complexity estimation remains one of the key problems in neural network research. The problem is further exacerbated in situations where the underlying system generator is non-stationary. In this paper, we introduce a novelty criterion for resource allocating networks (RANs) which is capable of being applied to both stationary and slowly varying non-stationary problems. The deficiencies of existing novelty criteria are discussed and the relative performances are demonstrated on two real-world problems : electricity load forecasting and exchange rate prediction.

    KW - feedforward neural nets

    KW - electricity load forecasting

    KW - exchange rate prediction

    KW - extended Kalman filter training

    KW - algorithm

    KW - network growth

    KW - network growth prescription

    KW - nonstationary real-world problems

    KW - novelty criterion

    KW - radial basis function network resource allocation

    KW - signal processing theory

    KW - slowly varying nonstationary environment

    UR - http://www.scopus.com/inward/record.url?scp=0030649838&partnerID=8YFLogxK

    UR - http://ieeexplore.ieee.org/servlet/opac?punumber=4811

    M3 - Chapter

    SN - 0852966903

    VL - 440

    T3 - Conference publication

    SP - 48

    EP - 52

    BT - Fifth International Conference on Artificial Neural Networks

    PB - IEEE

    ER -

    McLachlan A. An improved novelty criterion for resource allocating networks. In Fifth International Conference on Artificial Neural Networks. Vol. 440. IEEE. 1997. p. 48-52. (Conference publication).