Essec\Faculty\Model\Contribution {#2220
  #_index: "academ_contributions"
  #_id: "16507"
  #_source: array:26 [
    "id" => 16507
    "slug" => "16507-likelihood-free-inference-for-multivariate-generalized-pareto-models"
    "yearMonth" => "2026-05"
    "year" => 2026
    "title" => "Likelihood-Free Inference for Multivariate Generalized Pareto Models"
    "description" => "KRATZ, M., AKA, S. et NAVEAU, P. (2026). <i>Likelihood-Free Inference for Multivariate Generalized Pareto Models</i>. WP 2603, ESSEC Business School."
    "authors" => array:3 [
      0 => array:3 [
        "name" => "KRATZ Marie"
        "bid" => "B00072305"
        "slug" => "kratz-marie"
      ]
      1 => array:1 [
        "name" => "AKA Samira"
      ]
      2 => array:1 [
        "name" => "NAVEAU Philippe"
      ]
    ]
    "ouvrage" => ""
    "keywords" => array:3 [
      0 => "simulation-based inference"
      1 => "minimum distance estimation"
      2 => "multivariate extremes"
    ]
    "updatedAt" => "2026-06-02 13:24:05"
    "publicationUrl" => "https://hal.science/hal-05633438v1"
    "publicationInfo" => array:3 [
      "pages" => ""
      "volume" => ""
      "number" => ""
    ]
    "type" => array:2 [
      "fr" => "Documents de travail"
      "en" => "Working Papers"
    ]
    "support_type" => array:2 [
      "fr" => "Cahier de Recherche"
      "en" => "Working Papers"
    ]
    "countries" => array:2 [
      "fr" => null
      "en" => null
    ]
    "abstract" => array:2 [
      "fr" => """
        Likelihood-based inference for multivariate extreme-value models is often unreliable or infeasible when likelihoods are intractable or supports are discrete. This challenge is particularly acute for multivariate discrete generalized Pareto models, where both marginal tail behavior and dependence must be inferred from sparse exceedance samples. We propose a two-stage likelihood-free inference procedure, termed AW–\n
        NBE (Adaptive Wasserstein Neural Bayes Estimator), that combines neural Bayes estimation with a targeted optimal transport refinement step based on the Sinkhorn discrepancy. In the first stage, a neural Bayes estimator trained on simulated data provides fast and stable initial parameter estimates. In the second stage, these estimates are locally refined by minimizing the Sinkhorn divergence between the empirical\n
        distributions of observed and simulated exceedances. This refinement reduces the Sinkhorn discrepancy between the empirical distributions of observed and simulated exceedances, while preserving dependence features learned by the neural estimator.\n
        Model adequacy is assessed using new optimal transport based multivariate Q–Q and potential diagnostics. Applications to financial log-returns and Swiss dry spell exceedances suggest that AW–NBE can improve parameter inferences compared to estimation using solely, either the Sinkhorn discrepancy, or the standard neural Bayes estimators and censored likelihood estimation.
        """
      "en" => """
        Likelihood-based inference for multivariate extreme-value models is often unreliable or infeasible when likelihoods are intractable or supports are discrete. This challenge is particularly acute for multivariate discrete generalized Pareto models, where both marginal tail behavior and dependence must be inferred from sparse exceedance samples. We propose a two-stage likelihood-free inference procedure, termed AW–\n
        NBE (Adaptive Wasserstein Neural Bayes Estimator), that combines neural Bayes estimation with a targeted optimal transport refinement step based on the Sinkhorn discrepancy. In the first stage, a neural Bayes estimator trained on simulated data provides fast and stable initial parameter estimates. In the second stage, these estimates are locally refined by minimizing the Sinkhorn divergence between the empirical\n
        distributions of observed and simulated exceedances. This refinement reduces the Sinkhorn discrepancy between the empirical distributions of observed and simulated exceedances, while preserving dependence features learned by the neural estimator.\n
        Model adequacy is assessed using new optimal transport based multivariate Q–Q and potential diagnostics. Applications to financial log-returns and Swiss dry spell exceedances suggest that AW–NBE can improve parameter inferences compared to estimation using solely, either the Sinkhorn discrepancy, or the standard neural Bayes estimators and censored likelihood estimation.
        """
    ]
    "authors_fields" => array:2 [
      "fr" => "Systèmes d'Information, Data Analytics et Opérations"
      "en" => "Information Systems, Data Analytics and Operations"
    ]
    "indexedAt" => "2026-06-04T11:23:33.000Z"
    "docTitle" => "Likelihood-Free Inference for Multivariate Generalized Pareto Models"
    "docSurtitle" => "Working Papers"
    "authorNames" => "<a href="/cv/kratz-marie">KRATZ Marie</a>, AKA Samira, NAVEAU Philippe"
    "docDescription" => "<span class="document-property-authors">KRATZ Marie, AKA Samira, NAVEAU Philippe</span><br><span class="document-property-authors_fields">Information Systems, Data Analytics and Operations</span> | <span class="document-property-year">2026</span>"
    "keywordList" => "<a href="#">simulation-based inference</a>, <a href="#">minimum distance estimation</a>, <a href="#">multivariate extremes</a>"
    "docPreview" => "<b>Likelihood-Free Inference for Multivariate Generalized Pareto Models</b><br><span>2026-05 | Working Papers </span>"
    "docType" => "research"
    "publicationLink" => "<a href="https://hal.science/hal-05633438v1" target="_blank">Likelihood-Free Inference for Multivariate Generalized Pareto Models</a>"
  ]
  +lang: "en"
  +"_score": 8.502932
  +"_ignored": array:2 [
    0 => "abstract.en.keyword"
    1 => "abstract.fr.keyword"
  ]
  +"parent": null
}