Parameterized neural networks for high-energy physics

Pierre Baldi, Kyle Cranmer, Taylor Faucett, Peter Sadowski, Daniel Whiteson

    Research output: Contribution to journalArticle

    Abstract

    We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs to include not only measured features but also physics parameters. The physics parameters represent a smoothly varying learning task, and the resulting parameterized classifier can smoothly interpolate between them and replace sets of classifiers trained at individual values. This simplifies the training process and gives improved performance at intermediate values, even for complex problems requiring deep learning. Applications include tools parameterized in terms of theoretical model parameters, such as the mass of a particle, which allow for a single network to provide improved discrimination across a range of masses. This concept is simple to implement and allows for optimized interpolatable results.

    Original languageEnglish (US)
    Article number235
    JournalEuropean Physical Journal C
    Volume76
    Issue number5
    DOIs
    StatePublished - May 1 2016

    Fingerprint

    High energy physics
    classifiers
    Classifiers
    Neural networks
    learning
    physics
    Physics
    machine learning
    discrimination
    Learning systems
    energy
    education

    ASJC Scopus subject areas

    • Physics and Astronomy (miscellaneous)
    • Engineering (miscellaneous)

    Cite this

    Baldi, P., Cranmer, K., Faucett, T., Sadowski, P., & Whiteson, D. (2016). Parameterized neural networks for high-energy physics. European Physical Journal C, 76(5), [235]. https://doi.org/10.1140/epjc/s10052-016-4099-4

    Parameterized neural networks for high-energy physics. / Baldi, Pierre; Cranmer, Kyle; Faucett, Taylor; Sadowski, Peter; Whiteson, Daniel.

    In: European Physical Journal C, Vol. 76, No. 5, 235, 01.05.2016.

    Research output: Contribution to journalArticle

    Baldi, P, Cranmer, K, Faucett, T, Sadowski, P & Whiteson, D 2016, 'Parameterized neural networks for high-energy physics', European Physical Journal C, vol. 76, no. 5, 235. https://doi.org/10.1140/epjc/s10052-016-4099-4
    Baldi, Pierre ; Cranmer, Kyle ; Faucett, Taylor ; Sadowski, Peter ; Whiteson, Daniel. / Parameterized neural networks for high-energy physics. In: European Physical Journal C. 2016 ; Vol. 76, No. 5.
    @article{a21e667029a04113b58e757d4dd5fd2e,
    title = "Parameterized neural networks for high-energy physics",
    abstract = "We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs to include not only measured features but also physics parameters. The physics parameters represent a smoothly varying learning task, and the resulting parameterized classifier can smoothly interpolate between them and replace sets of classifiers trained at individual values. This simplifies the training process and gives improved performance at intermediate values, even for complex problems requiring deep learning. Applications include tools parameterized in terms of theoretical model parameters, such as the mass of a particle, which allow for a single network to provide improved discrimination across a range of masses. This concept is simple to implement and allows for optimized interpolatable results.",
    author = "Pierre Baldi and Kyle Cranmer and Taylor Faucett and Peter Sadowski and Daniel Whiteson",
    year = "2016",
    month = "5",
    day = "1",
    doi = "10.1140/epjc/s10052-016-4099-4",
    language = "English (US)",
    volume = "76",
    journal = "European Physical Journal C",
    issn = "1434-6044",
    publisher = "Springer New York",
    number = "5",

    }

    TY - JOUR

    T1 - Parameterized neural networks for high-energy physics

    AU - Baldi, Pierre

    AU - Cranmer, Kyle

    AU - Faucett, Taylor

    AU - Sadowski, Peter

    AU - Whiteson, Daniel

    PY - 2016/5/1

    Y1 - 2016/5/1

    N2 - We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs to include not only measured features but also physics parameters. The physics parameters represent a smoothly varying learning task, and the resulting parameterized classifier can smoothly interpolate between them and replace sets of classifiers trained at individual values. This simplifies the training process and gives improved performance at intermediate values, even for complex problems requiring deep learning. Applications include tools parameterized in terms of theoretical model parameters, such as the mass of a particle, which allow for a single network to provide improved discrimination across a range of masses. This concept is simple to implement and allows for optimized interpolatable results.

    AB - We investigate a new structure for machine learning classifiers built with neural networks and applied to problems in high-energy physics by expanding the inputs to include not only measured features but also physics parameters. The physics parameters represent a smoothly varying learning task, and the resulting parameterized classifier can smoothly interpolate between them and replace sets of classifiers trained at individual values. This simplifies the training process and gives improved performance at intermediate values, even for complex problems requiring deep learning. Applications include tools parameterized in terms of theoretical model parameters, such as the mass of a particle, which allow for a single network to provide improved discrimination across a range of masses. This concept is simple to implement and allows for optimized interpolatable results.

    UR - http://www.scopus.com/inward/record.url?scp=84964849143&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84964849143&partnerID=8YFLogxK

    U2 - 10.1140/epjc/s10052-016-4099-4

    DO - 10.1140/epjc/s10052-016-4099-4

    M3 - Article

    VL - 76

    JO - European Physical Journal C

    JF - European Physical Journal C

    SN - 1434-6044

    IS - 5

    M1 - 235

    ER -