Počet záznamů: 1  

Causal network discovery by iterative conditioning: Comparison of algorithms

  1. 1.
    SYSNO ASEP0520385
    Druh ASEPJ - Článek v odborném periodiku
    Zařazení RIVJ - Článek v odborném periodiku
    Poddruh JČlánek ve WOS
    NázevCausal network discovery by iterative conditioning: Comparison of algorithms
    Tvůrce(i) Kořenek, Jakub (UIVT-O) ORCID, RID, SAI
    Hlinka, Jaroslav (UIVT-O) RID, SAI, ORCID
    Číslo článku013117
    Zdroj.dok.Chaos - ISSN 1054-1500
    Roč. 30, č. 1 (2020)
    Poč.str.14 s.
    Jazyk dok.eng - angličtina
    Země vyd.US - Spojené státy americké
    Klíč. slovacausality inference ; complex networks ; transfer entropy ; conditional mutual information ; Granger causality
    Vědní obor RIVIN - Informatika
    Obor OECDComputer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
    CEPNV15-29835A GA MZd - Ministerstvo zdravotnictví
    NV15-33250A GA MZd - Ministerstvo zdravotnictví
    NV17-28427A GA MZd - Ministerstvo zdravotnictví
    GA19-16066S GA ČR - Grantová agentura ČR
    GA19-11753S GA ČR - Grantová agentura ČR
    Způsob publikováníOmezený přístup
    Institucionální podporaUIVT-O - RVO:67985807
    UT WOS000539636200007
    EID SCOPUS85078309242
    DOI10.1063/1.5115267
    AnotaceEstimating causal interactions in complex dynamical systems is an important problem encountered in many fields of current science. While a theoretical solution for detecting the causal interactions has been previously formulated in the framework of prediction improvement, it generally requires the computation of high-dimensional information functionals—a situation invoking the curse of dimensionality with increasing network size. Recently, several methods have been proposed to alleviate this problem, based on iterative procedures for the assessment of conditional (in)dependences. In the current work, we bring a comparison of several such prominent approaches. This is done both by theoretical comparison of the algorithms using a formulation in a common framework and by numerical simulations including realistic complex coupling patterns. The theoretical analysis highlights the key similarities and differences between the algorithms, hinting on their comparative strengths and weaknesses. The method assumptions and specific properties such as false positive control and order-dependence are discussed. Numerical simulations suggest that while the accuracy of most of the algorithms is almost indistinguishable, there are substantial differences in their computational demands, ranging theoretically from polynomial to exponential complexity and leading to substantial differences in computation time in realistic scenarios depending on the density and size of networks. Based on the analysis of the algorithms and numerical simulations, we propose a hybrid approach providing competitive accuracy with improved computational efficiency. Characterization of the structure of interactions in large heterogeneous systems based on observational data has become one of the dominant challenges across scientific fields. In many cases, measurements of the dynamical behavior are available, allowing inference of causal interactions among the subsystems by exploiting the principle of temporal precedence of the cause before the effect. Half a century ago, Sir Clive Granger proposed a formal treatment of the problem of detecting such interactions, based on statistical testing of the improvement of the prediction of a target variable by a candidate source variable. This process has been generalized to nonlinear processes using the framework of information theory. However, the practical applicability of this methodology has been hindered by the need to properly account for all other potential intervening variables in the system, bringing in both computational and accuracy issues growing with network size. In this work, we compare several prominent algorithms proposed recently for estimating causal structure in large networks. We introduce the algorithms within a common framework highlighting the similarities and differences and compare their accuracy and computational demands on both simulated random networks and realistic examples derived from real-world brain and climate dynamics datasets. Finally, we suggest an algorithm with competitive accuracy and faster performance.
    PracovištěÚstav informatiky
    KontaktTereza Šírová, sirova@cs.cas.cz, Tel.: 266 053 800
    Rok sběru2021
    Elektronická adresahttp://dx.doi.org/10.1063/1.5115267
Počet záznamů: 1