Articles producció científica> Enginyeria Química

Node Metadata Can Produce Predictability Crossovers in Network Inference Problems

  • Identification data

    Identifier: imarina:9243524
    Authors:
    Fajardo-Fontiveros, OscarGuimera, RogerSales-Pardo, Marta
    Abstract:
    Predicting whether two drugs have a harmful interaction or whether someone is going to like a certain movie are examples of network inference problems. In these problems, the goal is to predict new interactions (between drugs or between people and movies) based on some previously observed interactions. Having additional information about the network nodes or their metadata (for example, the mechanism of action of the drugs or the age of the individuals) helps to make better predictions, though it is not clear why or how. Here, we explore how that improvement happens.We study a very general network inference problem and show that node metadata do not affect the inference problem gradually. Rather, even when the importance assigned to the metadata increases smoothly, the inference process crosses over from a data-dominated regime to a metadata-dominated regime. These crossovers show some similarities to transitions driven by temperature, where one finds energy- and entropy-dominated regimes. Importantly, optimal inference is often encountered exactly at this crossover.Our study opens the door to better understanding the role of metadata in network inference problems and, more broadly, establishes further connections between general inference problems and physical concepts such as phase transitions.
  • Others:

    Author, as appears in the article.: Fajardo-Fontiveros, Oscar; Guimera, Roger; Sales-Pardo, Marta
    Department: Enginyeria Química
    URV's Author/s: Fajardo Fontiveros, Oscar / Guimera Manrique, Roger / Sales Pardo, Marta
    Keywords: Mixed-membership prediction information-theory
    Abstract: Predicting whether two drugs have a harmful interaction or whether someone is going to like a certain movie are examples of network inference problems. In these problems, the goal is to predict new interactions (between drugs or between people and movies) based on some previously observed interactions. Having additional information about the network nodes or their metadata (for example, the mechanism of action of the drugs or the age of the individuals) helps to make better predictions, though it is not clear why or how. Here, we explore how that improvement happens.We study a very general network inference problem and show that node metadata do not affect the inference problem gradually. Rather, even when the importance assigned to the metadata increases smoothly, the inference process crosses over from a data-dominated regime to a metadata-dominated regime. These crossovers show some similarities to transitions driven by temperature, where one finds energy- and entropy-dominated regimes. Importantly, optimal inference is often encountered exactly at this crossover.Our study opens the door to better understanding the role of metadata in network inference problems and, more broadly, establishes further connections between general inference problems and physical concepts such as phase transitions.
    Thematic Areas: Physics, multidisciplinary Physics and astronomy (miscellaneous) Physics and astronomy (all) Matemática / probabilidade e estatística General physics and astronomy Engenharias iv Astronomia / física
    licence for use: https://creativecommons.org/licenses/by/3.0/es/
    Author's mail: roger.guimera@urv.cat oscar.fajardo@estudiants.urv.cat oscar.fajardo@estudiants.urv.cat marta.sales@urv.cat
    Author identifier: 0000-0002-3597-4310 0000-0002-8140-6525
    Record's date: 2024-10-19
    Papper version: info:eu-repo/semantics/publishedVersion
    Link to the original source: https://journals.aps.org/prx/abstract/10.1103/PhysRevX.12.011010
    Licence document URL: https://repositori.urv.cat/ca/proteccio-de-dades/
    Papper original source: Physical Review x. 12 (1): 011010-
    APA: Fajardo-Fontiveros, Oscar; Guimera, Roger; Sales-Pardo, Marta (2022). Node Metadata Can Produce Predictability Crossovers in Network Inference Problems. Physical Review x, 12(1), 011010-. DOI: 10.1103/physrevx.12.011010
    Article's DOI: 10.1103/physrevx.12.011010
    Entity: Universitat Rovira i Virgili
    Journal publication year: 2022
    Publication Type: Journal Publications
  • Keywords:

    Physics and Astronomy (Miscellaneous),Physics, Multidisciplinary
    Mixed-membership
    prediction
    information-theory
    Physics, multidisciplinary
    Physics and astronomy (miscellaneous)
    Physics and astronomy (all)
    Matemática / probabilidade e estatística
    General physics and astronomy
    Engenharias iv
    Astronomia / física
  • Documents:

  • Cerca a google

    Search to google scholar