tag:blogger.com,1999:blog-31964354054227188702024-03-26T03:04:59.943-07:00UQSayUQ, DACE and related topics @ Paris SaclayJulien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comBlogger73125tag:blogger.com,1999:blog-3196435405422718870.post-50369673006152401192024-03-14T01:34:00.000-07:002024-03-26T03:04:27.271-07:00UQSay #71<div style="text-align: justify;">
<p>The seventy-first UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, March 21, 2024.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay71_poster.pdf">
<img border="0"
data-original-height="708"
d1ata-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgFZwqpPMhSuHbNFsV8Lyk1guGDJuazdvPwM1eBFEPv2ty4_LF8IjM59a4vvyelH7ffxbWE-APTMP3CUKLC6Dpl6xSjSvwoWb0mLGPmZ71da6wZZ-2W7LuFMcUKTll47fFS6kgL4BEKd0SWjUaL8Eo_5uzvjEqFHufOUxYFJxRLW9m68tjzRaUUupuUD5Gc/s600/uqsay71_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.ifpenergiesnouvelles.fr/samourai">Morgane Menz</a> (<a href="https://www.ifpenergiesnouvelles.fr/tags/mathematiques-et-informatique">IFPEN</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay71_slides_mmenz.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Archissur: an Active Recovery of Constrained and Hidden Sets by SUR method. Coupling with adaptive space-filling and optimization</h4>
<p>The analysis of simulated engineering systems (robust optimization, reliability assessment, …) generally requires numerous computationally expensive code simulations with different possible sets of values of design and environmental input variables. However, the simulators can encounter simulation crashes due to convergence issues for some values of both input variables. These failures correspond to a hidden constraint and might be as costly to evaluate as a feasible simulation. The presence of such crashes must be managed in a wise way, in order to target feasible input areas and thus avoid unnecessary irrelevant simulations. </p>
<p> In this context, we propose an adaptive strategy to learn the hidden constraint at a reduced numerical cost based only on a limited number of binary observations corresponding to failure or non-failure status. Our approach is a Gaussian Process Classifier active learning method based on Stepwise Uncertainty Reduction strategies to assess hidden constraints prediction. A numerically effective formulation of the enrichment criterion suited for classification is provided.
Additionally, the proposed enrichment criterion is employed to address metamodeling and optimization in the presence of hidden constraints..</p>
<p>Reference: <a href="https://hal.science/hal-03848238/">Estimation of simulation failure set with active learning based on Gaussian Process classifiers and random set theory</a>, 2023.</p>
<p>Joint work with <a href="https://www.ifpenergiesnouvelles.fr/node/900">Miguel Munoz-Zuniga </a>(IFPEN) & <a href="https://www.ifpenergiesnouvelles.fr/page/delphine-sinoquet">Delphine Sinoquet </a>(IFPEN).</p>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-4421478067425200782024-02-28T07:10:00.000-08:002024-03-26T03:03:50.027-07:00UQSay #70<div style="text-align: justify;">
<p>The seventieth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, March 7, 2024.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay70_poster.pdf">
<img border="0"
data-original-height="708"
d1ata-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjp5sjS-thFu0gYh8n3rlaY_Pr5hxw65T17eGFAqwH8NKglmTyr4ze8YJTcqTZoNZ-IhKuNVXL492kG8Lbu47NJtLWwkkhY0S7nALpstHdlnGb8VbQlHasrK-F3DhwvePR1ikZvW-tqty7Vv__GSbsPj2kFTtvvjOw8bZnNn2aHhoWyTPuSuCJRslgGm_1l/s708/uqsay70_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://ehbihenscoding.github.io/kerleguer/index.html">Baptiste Kerleguer</a> (<a href="https://www-dam.cea.fr/dam/">CEA, DAM/DIF</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay70_slides_bkerleguer.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">A Bayesian neural network approach to multi-fidelity surrogate
modeling</h4>
<p>This talk deals with the surrogate modeling of computer code results that can be evaluated at different levels of accuracy and computational
cost, called multi-fidelity. We propose a method combining Gaussian process (GP) regression on low-fidelity data and a Bayesian neural
network (BNN) on high-fidelity data. The novelty, compared with the state of the art, is that uncertainties are taken into account
at all fidelity levels. The prediction uncertainty of the low-fidelity level is transmitted by Gauss-Hermite quadrature to the high-fidelity
level. In addition, this method takes into account non-nested designs of experiment and non-linear interactions between levels. The proposed
approach is then compared to several multi-fidelity GP regression methods on analytic functions and on a computer code.</p>
<p>Reference: <a href="https://arxiv.org/abs/2312.02575">A Bayesian neural network approach to multi-fidelity surrogate modeling</a>, IJUQ 14.1, 2024.</p>
<p>Joint work with <a href="https://josselin-garnier.org/">Josselin Garnier </a>(CMAP) & <a href="https://www.linkedin.com/in/claire-cannamela-273b7511">Claire Cannamela </a>(CEA, DAM/DIF).</p>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-72072708597379490372024-02-01T09:29:00.000-08:002024-03-26T03:03:33.993-07:00UQSay #69<div style="text-align: justify;">
<p>The sixty-ninth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, February 8, 2024.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay69_poster.pdf">
<img border="0"
data-original-height="708"
d1ata-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEi2CiSF_aFTubMoRrl5T6axymFwuoWWFxIxCDFE_EMLTxDdty7HOEIAomLmHZTlVTRPJmXlE7l7bxjBgfUrbpNLxFu8pxjyT8QJKoBM_nBcVLZxMUCq1rg0LHS0zJXBOAgYDZt7kqoUcoicewjFremiC4BnPLaOh2kaj7d8u9GyWQHEUUglqCCfWBhgqVTx/s600/uqsay69_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://marouaneilidrissi.com">Marouane Il Idrissi</a> (<a href="https://www.edf.fr/groupe-edf/inventer-l-avenir-de-l-energie/r-d-un-savoir-faire-mondial">EDF R&D</a> - <a href="https://www.math.univ-toulouse.fr/">IMT</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay69_slides_milidrissi.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Generalized Hoeffding decomposition and the linear nature of non-linearity</h4>
<p>Hoeffding’s decomposition of random outputs traditionally requires the inputs to be mutually independent. It allows uniquely decomposing a square-integrable function as a sum taken over every subset of inputs. Generalizing this result to non-mutually independent inputs has been a recent challenge in the literature on sensitivity analysis. Proposed solutions exist, but they require relatively restrictive assumptions on the distribution of the inputs. However, Hoeffding’s decomposition can be generalized under two reasonable assumptions on the inputs’ distribution: non-perfect functional dependence and non-degenerate stochastic dependence. </p>
<p> This generalization requires approaching the problem using a framework at the cornerstone of probability theory, functional analysis, and combinatorics. From this perspective, it can be seen as finding a direct-sum decomposition of a particular Lebesgue space, unveiling a surprisingly linear approach to handling stochastic and functional non-linearities. The proposed "ortho-canonical decomposition" relies on oblique projections rather than the traditional conditional expectations.
Ultimately, it allows the definition of intuitive and interpretable sensitivity indices, which offers a path toward a more precise uncertainty quantification.</p>
<p>In this talk, we will delve into the unconventional framework used, discuss its nuances, and explore the various perspectives and challenges it offers.</p>
<p>Reference: <a href="https://arxiv.org/abs/2310.06567">Understanding black-box models with dependent inputs through a generalization of Hoeffding's decomposition</a>, 2023 [<a href="https://github.com/milidris/GeneralizedAnova">github</a>].</p>
<p>Joint work with <a href="https://perso.lpsm.paris/~bousquet/">Nicolas Bousquet </a>(EDF R&D - LPSM), <a href="https://www.math.univ-toulouse.fr/~gamboa/">Fabrice Gamboa </a>(IMT), <a href="https://biooss1.wixsite.com/bertrand">Bertrand Ioss </a>(EDF R&D - IMT) and <a href="https://perso.math.univ-toulouse.fr/loubes/">Jean-Michel Loubes </a>(IMT).</p>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-87664360261216026592024-01-18T06:52:00.000-08:002024-03-26T03:03:13.845-07:00UQSay #68<div style="text-align: justify;">
<p>The sixty-eighth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, January 25, 2024.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay68_poster.pdf">
<img border="0"
data-original-height="708"
d1ata-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjYh-e6XlnFJZlTV31nzwV3JkytFrGnMFItoGPG04_VVdzt79MH5FfmsgMVJRfDVi_zB0wpLBoQF3gp9VeeM4PZpO_7VgLQBMNDwGuxkenhR7RcfvOIwtqsUWuFzE-DYFuMSxgrNEa-evK5-314s0l1fd81LnOY0MXCfq4tuyzQLr2J0B21KfqqNpxP1R-P/s708/uqsay68_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.researchgate.net/profile/Gabriel-Sarazin-2">Gabriel Sarazin</a> (<a href="https://www.cea.fr/paris-saclay/Pages/Accueil.aspx">CEA Paris Saclay</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay68_slides_gsarazin.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Towards more interpretable kernel-based sensitivity analysis</h4>
<p>When working with a computationally-expensive simulation code involving a large number of uncertain physical parameters, it is often advisable to perform a preliminary sensitivity analysis in order to identify which input variables will really be useful for surrogate modelling. On paper, the total-order Sobol' indices fulfill this role perfectly, since they are able to detect any type of input-output dependence, while being interpretable as simple percentages of the output variance. However, in many situations, their accurate estimation remains a thorny issue, despite remarkable progress in that direction over the past few months. In this context where inference is strongly constrained, kernel methods have emerged as an excellent alternative, notably through the Hilbert-Schmidt independence criterion (HSIC). Although they offer undeniable advantages over Sobol' indices, HSIC indices are much harder to understand, and this lack of interpretability is a major obstacle to their wider dissemination. In order to marry the advantages of Sobol' and HSIC indices, an ANOVA-like decomposition allows to define HSIC-ANOVA indices at all orders, just as would be done for Sobol' indices. This recent contribution is the starting point of this presentation.</p>
<p>The main objective of this talk is to provide deeper insights into the HSIC-ANOVA framework. One major difference with the basic HSIC framework lies in the use of specific input kernels (like Sobolev kernels). First, a dive into the universe of cross-covariance operators will allow to better understand how sensitivity is measured by HSIC-ANOVA indices, and what type of input-output dependence is captured by each term of the HSIC-ANOVA decomposition. Then, a brief study of Sobolev kernels, focusing more particularly on their feature maps, will reveal what kind of simulators are likely to elicit HSIC-ANOVA interactions. It will also be demonstrated that Sobolev kernels are characteristic, which ensures that HSIC-ANOVA indices can be used to test input-output independence. Finally, a test procedure will be proposed for the total-order HSIC-ANOVA index, and it will be shown (numerically) that the resulting test of independence is at least as powerful as the standard test (based on two Gaussian kernels).</p>
<p>Reference: <a href="https://cea.hal.science/cea-04320711/document">New insights into the feature maps of Sobolev kernels: application in global sensitivity analysis</a>, 2023 [<a href="https://rdrr.io/cran/sensitivity/man/sensiHSIC.html">sensiHSIC</a> & <a href="https://rdrr.io/cran/sensitivity/man/testHSIC.html"> testHSIC </a> in R package sensitivity].</p>
<p>Joint work with <a href="https://scholar.google.com/citations?user=fYND7JQAAAAJ&hl=en">Amandine Marrel </a>(CEA Cadarache), <a href="https://scholar.google.fr/citations?user=3RqZWKsAAAAJ&hl=fr">Sébastien Da Veiga </a>(ENSAI) and <a href="https://scholar.google.com/citations?user=xBfXuv0AAAAJ&hl=en">Vincent Chabridon </a>(EDF R&D).</p>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-81190283421199902782023-12-06T02:41:00.000-08:002023-12-18T01:02:43.471-08:00UQSay #67<div style="text-align: justify;">
<p>The sixty-seventh UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, December 14, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay67_poster.pdf">
<img border="0"
data-original-height="708"
d1ata-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgDq7EzxKHbsRHVEs8MxweJ41FvT7SnijbQ60XCqI5YqDgGhnS42kZGrtbqIRukDz530RgIjALV27s-8Y5Zo29Mp147GS0BsZHu7JzwXfxJ6XRY5e9d2bbqzt6fwdZAfNAN7VVSMJ_adO34dXDt1MXZ51rN97x-s8GHW35znPuFwzUBYFfEAn1ZapZkmcpB/s600/uqsay67_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.imo.universite-paris-saclay.fr/fr/perso/pierre-humbert/">Pierre Humbert</a> (<a href="https://www.imo.universite-paris-saclay.fr/fr/">LMO</a> - <a href="https://team.inria.fr/celeste/">INRIA</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay67_slides_phumbert.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">One-Shot Federated Conformal Prediction</h4>
<p>In this presentation, we will focus on a method for constructing prediction sets in a federated learning setting where only one round of communication between the agents and the server is allowed (one-shot).
More precisely, by defining a particular estimator called the quantile-of-quantiles, we will prove that for any distribution, it is possible to produce marginally (and training-conditionally) valid prediction sets.
Over a wide range of experiments, we will show that we are able to obtain prediction sets whose coverage and length are very similar to those obtained in a centralized setting, making our method particularly
well-suited to perform conformal predictions in a one-shot federated learning setting.</p>
<p>Reference: <a href="https://proceedings.mlr.press/v202/humbert23a.html">One-Shot Federated Conformal Prediction</a>, ICML 2023</p>
<p>Joint work with <a href="https://batistelb.github.io/">Batiste Le Bars</a>, <a href="http://researchers.lille.inria.fr/abellet/">Aurélien Bellet</a> and <a href="https://www.imo.universite-paris-saclay.fr/~sylvain.arlot/">Sylvain Arlot</a>.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-58330525779002682772023-11-24T01:33:00.000-08:002023-12-18T00:50:21.257-08:00UQSay #66<p>The sixty-sixth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, November 30, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay66_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjoaaP9tUiJfYExgmBS9czCIfgYh-QR3Ye6evjGnqWyebQeouE9T3hCsGNxk0MR5amOSSBjWYUu1o5pxY2LquFJm92vqdkcGNDU7gBomM9ZfZLbpRE346MsHTFZx2zY9vs9npte2MwCPRmfge3nfl1ReRKo78XapwEflMG_HOQCxsAtUgBtseCxWfmyO3XA/s708/uqsay66_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.marquette.edu/mathematical-and-statistical-sciences/directory/elaine-spiller.php">Elaine Spiller</a> (<a href="https://www.marquette.edu/arts-sciences/">Marquette University</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay66_slides_espiller.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Two recent advances in UQ with Gaussian process models:<br> the zGP and the PPLE</h4>
<p>Gaussian processes (GPs) are an effective and widely used tools to emulate computer simulations of physical process models for uncertainty quantification (UQ). Over the last 10-15 years, GP modeling of computer simulations has advanced tremendously to handle challenges posed by complex and realistic simulators. We will discuss two recent challenges.
The first challenge is the "zero-problem” — simulations that result in positive, real-valued output or zero. Such zero-censored data pose a significant obstacle to GP emulators because of both the inherent non-stationary and because GPs have full support. The second challenge we will explore is emulating high-dimensional multi-physics simulations. Here we will combine two recent GP approaches: linked GP emulation (for coupled physical simulations) and parallel partial emulators (PPEs) for emulating simulators with high-dimensional output. The resulting parallel partial linked GP emulator (PPLE) proves an efficient approach to emulate high-dimensional multi-physics simulators.</p>
<p>Reference: <a>E.T. Spiller, R.W. Wolpert, P. Tierz & T.G. Asher, “<a href="https://epubs.siam.org/doi/10.1137/21M1467420">The zero problem: Gaussian process emulators for range constrained computer models</a>”, 2023. [<a href="https://github.com/SideofMan/zGP">github</a>]</p>
<p>Joint work with Robert Wolpert (<a href="https://www2.stat.duke.edu/~rlw/">Duke Univ.</a>) and Sue Minkoff (<a href="https://personal.utdallas.edu/~sminkoff/">Univ. of Texas</a>)
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-17216115898089061392023-11-05T05:06:00.000-08:002023-11-05T05:06:36.124-08:00UQSay #65<div style="text-align: justify;">
<p>The sixty-fifth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, November 16, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay65_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEj7MkYYbIL2ZxzuvGjwZ5l2z1Yg-7rpPeJ299SECxuhR9PcqszKalk_-pDGUNnW_Q7TkgwMqAQZ_6ywfk8UecJq1r-yiSATE6K2xr1-DbjJM2taAt4nY9HUMxyqEvo1PI4LrVBfLeFzbtLxiw2XweyZ0BuXUhyfviRxCZbzpKf546ePzYc_hpFlHqZ4fze_/s16000/uqsay65_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://people.eecs.berkeley.edu/~jordan/">Michael I. Jordan</a> (<a href="https://www.berkeley.edu/">Berkeley, University of California.</a>)</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Prediction-Powered Inference</h4>
<p>I introduce prediction-powered inference – a framework for performing valid statistical inference when an experimental data set is supplemented with predictions from a machine-learning system. Our framework yields provably valid conclusions without making any assumptions on the machine-learning algorithm that supplies the predictions. Higher accuracy of the predictions translates to smaller confidence intervals, permitting more powerful inference. Prediction-powered inference yields simple algorithms for computing valid confidence intervals for statistical objects such as means, quantiles, and linear and logistic regression coefficients. I demonstrate the benefits of prediction-powered inference with data sets from proteomics, genomics, electronic voting, remote sensing, census analysis, and ecology.</p>
<p>Reference:
<a>A. Angelopoulos, S. Bates, C. Fannjiang, M. I. Jordan, T. Zrnic, “<a href="https://arxiv.org/abs/2301.09633">Prediction-Powered Inference</a>”, 2023</a>.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-51102494000010532023-10-30T06:54:00.003-07:002023-11-12T13:24:53.689-08:00UQSay #64<p>The sixty-fourth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, November 2, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay64_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhMaMcRPRIO_mtCTpXdReQtuIxshp1eKTelAH15JXPPiE9THa0-Gx5NwgErj6a_poOYR2EVME0DaAzJO3gDlRsIrVJIRgO4q86ml2iUJtlegbWm96yDRBUZmBYUkN8x1nhNKC5dhnDXUsCEgeZusiFUDv4YydTzqjL8G9CCfvd96XO_c87cYWVmLKtZggfR/s16000/uqsay64_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="http://christophmolnar.com/">Christoph Molnar</a> & <a href="https://freieslebent.github.io/"> Timo Freiesleben</a> (<a href="https://uni-tuebingen.de/en/research/core-research/cluster-of-excellence-machine-learning">Machine Learning in Science Cluster, University of Tübingen.</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay64_slides_cmolnar_tfreiesleben.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Supervised Machine Learning in Science</h4>
<p>From folding proteins and predicting tornadoes to studying human nature — machine learning has changed science. Science always had an intimate relationship with prediction, but machine learning intensifies this focus. Can this hyper-focus on prediction models be justified? Can a machine learning model be part of a scientific model? Or are we on the wrong track? We explore and justify the use of supervised machine learning in science. However, a pure and naive application of supervised learning won't get you far, because raw machine learning has so many insufficiencies that make it unusable in this form for science. Unintelligible models, lack of uncertainty quantification, lack of causality. But we already have all the puzzle pieces to fix machine learning, from incorporating domain knowledge and assuring the representativeness of the training data to robust, interpretable, and causal models. We bring together the philosophical justification and the solutions that make supervised machine learning a powerful tool for science.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-53420962233768844232023-10-14T12:16:00.001-07:002023-10-14T12:18:13.820-07:00UQSay #63<div style="text-align: justify;">
<p>The sixty-third UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, October 19, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay63_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhXsNjZfX_S8kWpPgw5vPUqvXq2zLey4LSNBik6evSuTM6aYj4C8E0BWTlMrfk6yvREcOyiffCLeAouSFjcYRFArqFR71MrOscmPymVuceYR_GPWoebOY1miFj5jBMJhGDNwIjyn9Fdi_HPsXcxG8Zzo_31bMprQpKIFL2u4I71ri1p2f4t2YpAgnqrF4yd/s16000/uqsay63_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.f6s.com/member/stefania-fresca">Stefania Fresca</a> (<a href="https://mox.polimi.it/">MOX, Dept. of Mathematics, Politecnico di Milano</a>)</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Deep learning-based reduced order models for the real-time approximation of parametrized PDEs</h4>
<p>Conventional reduced order models (ROMs) anchored to the assumption of modal linear superimposition, such as proper orthogonal decomposition (POD), may reveal inefficient when dealing with nonlinear
time-dependent parametrized PDEs, especially for problems featuring coherent structures propagating over time. To enhance ROM efficiency, we propose a nonlinear approach to set ROMs by exploiting
deep learning (DL) algorithms, such as convolutional neural networks. In the resulting DL-ROM, both the nonlinear trial manifold and the nonlinear reduced dynamics are learned in a non-intrusive way by
relying on DL algorithms trained on a set of full order model (FOM) snapshots, obtained for different parameter values. Performing then a former dimensionality reduction on FOM snapshots through
POD enables, when dealing with large-scale FOMs, to speedup training times, and decrease the network complexity, substantially. Accuracy and efficiency of the DL-ROM technique are assessed on different
parametrized PDE problems in cardiac electrophysiology, computational mechanics and fluid dynamics, possibly accounting for fluid-structure interaction (FSI) effects, where new queries to the DL-ROM can
be computed in real-time. Moreover, numerical results obtained by the application of DL-ROMs to the solution of an industrial application, i.e. the approximation of the structural or the electromechanical behaviour of Micro-Electro-Mechanical Systems (MEMS), will be shown.</p>
<p>References:</p>
<ol>
<li> G. Gobat, S. Fresca, A. Manzoni, A. Frangi, “<a href="https://doi.org/10.3390/s23063001">Reduced order modelling of nonlinear vibrating multiphysics microstructures with deep learning-based approaches</a>”, Sensors, vol. 23, no. 6, pp. 3001, 2023 .</li>
<li> S. Fresca, G. Gobat, P. Fedeli, A. Frangi, A. Manzoni, “<a href="https://arxiv.org/abs/2111.12511">Deep learning-based reduced order models for the real-time simulation of the nonlinear dynamics of microstructures</a>”, International Journal for Numerical Methods in Engineering, vol. 123, no. 20, pp. 4749-4777, 2022 .</li>
<li> S. Fresca, A. Manzoni, “<a href="https://arxiv.org/abs/2101.11845">POD-DL-ROM: enhancing deep learning-based reduced order models for nonlinear parametrized PDEs by proper orthogonal decomposition</a>”, Computer Methods in Applied Mechanics and Engineering, vol. 388, pp. 114181, 2022 .</li>
<li> S. Fresca, A. Manzoni, L. Dede’, A. Quarteroni, “<a href="https://doi.org/10.3389/fphys.2021.679076">POD-enhanced deep learning-based reduced order models for the real-time simulation of cardiac electrophysiology in the left atrium</a>”, Frontiers in Physiology, vol. 12, pp. 1431, 2021 .</li>
<li> S. Fresca, A. Manzoni, “<a href="https://arxiv.org/abs/2106.05722">Real-time simulation of parameter-dependent fluid flows through deep learning-based reduced order models</a>”, Fluids, vol. 6, no. 7, pp. 259, 2021 .</li>
<li> S. Fresca, A. Manzoni, L. Dede’, “<a href="https://doi.org/10.3389/fphys.2021.679076">A comprehensive deep learning-based approach to reduced order modeling of nonlinear time-dependent parametrized PDEs</a>”, Journal of Scientific Computing, vol. 87, no. 2, pp. 1-36, 2021 .</li>
</ol>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-25231037715904025732023-09-29T07:05:00.002-07:002023-11-12T13:14:47.566-08:00UQSay #62<div style="text-align: justify;">
<p>The sixty-second UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place online on Thursday afternoon, October 5, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay62_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEig2JEGgRVkgaGsmt1F3AP4auvyeGnAFyTVFhzU4UAzBNoEoh5HdLQsWJLV64MPCeMWFyuNkA69PD91ymlYPtUKpb3UQrjChRHLp6JydtGBxexdG5K6otmT3-UfJFGTnThmIETANKzgPGhBCvkomy3PuUsGBaS3-kq6rRoGJfv_U04mte-doE3AkESN_sPy/s16000/uqsay62_poster.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.imperial.ac.uk/people/sibo.cheng">Sibo Cheng</a> (<a href="https://www.imperial.ac.uk/data-science/">Data Science Inst.</a>, <a href="https://www.imperial.ac.uk/computing/">Imperial College London</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay62_slides_scheng.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Machine learning and data assimilation for high dimensional dynamical systems</h4>
<p>Data Assimilation (DA) and Machine Learning (ML) methods are extensively used in predicting and updating high-dimensional spatial-temporal dynamics. Typical applications span from computational fluid dynamics to geoscience and climate systems. In recent years, much effort has been given in combining DA and ML techniques with objectives including but not limited to dynamical system identification, reduced order surrogate modelling, error covariance specification and model error correction. This talk will provide an overview of state-of-the-art research in this interdisciplinary field, covering a wide range of applications. I will also present my unpublished work regarding efficient deep data assimilation with sparse observations and time-varying sensors. The proposed method, incorporating a deep learning inverse operator based on Voronoi tessellation into the assimilation objective function, is adept at handling sparse, unstructured, and time-varying sensor data.</p>
<p>Reference: S. Cheng, C. Quilodran-Casas, S. Ouala, A. Farchi, C. Liu, P. Tandeo, R. Fablet, D. Lucor, B. Iooss, J. Brajard, D. Xiao, T. Janjic, W. Ding, Y. Guo, A. Carrassi, M. Bocquet and R. Arcucci, “<a href="https://arxiv.org/abs/2303.10462">Machine Learning With Data Assimilation and Uncertainty Quantification for Dynamical Systems: A Review</a>”, IEEE/CAA Journal of Automatica Sinica, vol. 10, no. 6, pp. 1361–1387, June 2023.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-71682736846574591972023-06-03T04:47:00.006-07:002023-06-28T07:42:46.209-07:00UQSay #61<div style="text-align: justify;">
<p>The sixty-first UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, June 8, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay61_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEiE_mL_-2uEQoXoIpzJa0duwqjZhXZ3MLFRDHFonYsSdBKCdXCgINhAUHQmR7RiqQns84hkTZqZczlInEXdttMAngFCUgf534SPiOeIOYrq0l64A87wqXkXHe-NitGd6K_78p1Gm_H9SnQopUcJept8mdUZFMxgJt3Ex0A-gmRw0kM6u0y_ooNhg6IMlA/s708/uqsay61_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — Sophie Ricci</a> (<a href="https://cerfacs.fr/umr-5318-climat-environnement-couplages-et-incertitudes-ceci/">CECI, CERFACS & UMR 5318</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay60_slides_sfortunati.pdf">slides</a>]</h4>
<br />
<h4 style="text-align: left; font-weight: bold;">On the merits of using remote sensing Earth Observation data to reduce uncertainties in flood forecasting with ensemble-based data assimilation</h4>
<p>Flood simulation and forecast capability have been greatly improved thanks to advances in data assimilation (DA) strategies incorporating various types of observations; many are derived from spatial Earth Observation. This paper focuses on the assimilation of 2D flood observations derived from Synthetic Aperture Radar (SAR) images acquired during a flood event with a dual state-parameter Ensemble Kalman Filter (EnKF). Binary wet/dry maps are here expressed in terms of wet surface ratios (WSR) over a number of subdomains of the floodplain. This ratio is further assimilated jointly with in-situ water-level observations to improve the flow dynamics within the floodplain. However, the non-Gaussianity of the observation errors associated with SAR-derived measurements break a major hypothesis for the application of the EnKF, thus jeopardizing the optimality of the filter analysis. The novelty of this paper lies in the treatment of the non-Gaussianity of the SAR-derived WSR observations with a Gaussian anamorphosis process (GA). This DA strategy was validated and applied over the Garonne Marmandaise catchment (South-west of France) represented with the TELEMAC-2D hydrodynamic model, first in a twin experiment and then for a major flood event that occurred in January-February 2021. It was shown that assimilating SAR-derived WSR observations, in complement to the in-situ water-level observations significantly improves the representation of the flood dynamics. Also, the GA transformation brings further improvement to the DA analysis, while not being a critical component in the DA strategy. This study heralds a reliable solution for flood forecasting over poorly gauged catchments thanks to available remote-sensing datasets.</p>
<p>Joint work with T. H. Nguyen & A. Piacentini (CECI, CERFACS), E. Simon (INP, IRIT), R. Rodriguez-Suquet & S. Peña-Luque (CNES).</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-23854521642577668672023-05-17T00:57:00.010-07:002023-05-25T06:31:38.273-07:00UQSay #60<div style="text-align: justify;">
<p>The sixtieth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, May 25, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay60_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgRfb7iAZeow1jiSdfX8ZK7-aHcWVtu8nwsQ4Z3aQuYaNBSVvPVKqgMTEzFubXFxaYMbWqt9GHWWCKrHfs0qX3W42HmnhpEjMFHmG8IKc8c2fdj20u4MxQyuAWBo86f0rGtF3TfaPodVYLo1AUarUsFPNInWNLexvz3DI9FLCWrMMPkTMhKohlv_LDvCQ/s708/uqsay60_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://l2s.centralesupelec.fr/u/fortunati-stefano/">Stefano Fortunati</a> (<a href="https://l2s.centralesupelec.fr/">LSS</a> & <a href="https://www.ipsa.fr/en/">IPSA</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay60_slides_sfortunati.pdf">slides</a>]</h4>
<br />
<h4 style="text-align: left; font-weight: bold;">Matched, mismatched and semiparametric inference in elliptical distributions</h4>
<p>Any scientific experiment, which aims to gain some knowledge about a real-word phenomenon, starts with the data collection. In statistics, all the available knowledge about a phenomenon of interest is summarized in the probability density function (pdf) of the collected observations. To this end, we define a model as the family of pdfs that are able to statistically characterize the observations. The most used classes of models are the parametric ones which however require the perfect match between the actual data distribution and the assumed model itself. Nevertheless, in practice, a certain amount of mismatch is often inevitable. Therefore, being aware about the possible performance loss that the derived estimator could undergone under model misspecification is of crucial importance. Even more important would be the possibility to overcome this misspecification problem. This can be achieved by adopting the more general semiparametric characterization of the statistical behavior of the collected data. In this seminar we use the set of elliptical distribution as “fil rouge” to analyse the three above-mentioned aspects.</p>
<p>Joint work with F. Gini & M. S. Greco (University of Pisa, Italy), C. D. Richmond (Duke University, USA), A. M. Zoubir (TU Darmstadt, Germany), A. Renaux (L2S/UPS), F. Pascal (L2S/CentraleSupélec).</p>
<p>References:</p>
<ol>
<li>S. Fortunati, F. Gini, M. S. Greco and C. D. Richmond, “<a href="https://arxiv.org/abs/1709.08210">Performance Bounds for Parameter Estimation under Misspecified Models: Fundamental Findings and Applications</a>”, IEEE Signal Processing Magazine, vol. 34, no. 6, pp. 142-157, Nov. 2017.</li>
<li>S. Fortunati, F. Gini, M. S. Greco, A. M. Zoubir and and M. Rangaswamy, “<a href="https://arxiv.org/abs/1807.07811">Semiparametric Inference and Lower Bounds for Real Elliptically Symmetric Distributions</a>”, IEEE Transactions on Signal Processing, vol. 67, no. 1, pp. 164-177, 1 Jan.1, 2019.</li>
<li>S. Fortunati, F. Gini, M. S. Greco, A. M. Zoubir and and M. Rangaswamy, “<a href="https://arxiv.org/abs/1902.09541">Semiparametric CRB and Slepian-Bangs Formulas for Complex Elliptically Symmetric Distributions</a>”, IEEE Transactions on Signal Processing, vol. 67, no. 20, pp. 5352-5364, 15 Oct.15, 2019.</li>
<li>S. Fortunati, A. Renaux, F. Pascal, “<a href="https://arxiv.org/abs/2002.02239">Robust semiparametric efficient estimators in complex elliptically symmetric distributions</a>”, IEEE Transactions on Signal Processing, vol. 68, pp. 5003-5015, 2020.</li>
</ol>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Sidonie Lefebvrehttp://www.blogger.com/profile/15572376857277026896noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-65852337369005650742023-05-05T01:11:00.003-07:002023-05-12T02:31:45.288-07:00UQSay #59<div style="text-align: justify;">
<p>The fifty-nineth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, May 11, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay59_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgMXhC2mcE9LJ4SY3_6IxmG2r5d5KNZWkBs2kpi9H5P9eeO2ui1RDIbpzHzXNENHdr7iyuYUiPgVZGlqbvfBxOo9Utz7Wi0gVE_CaHgBsEvpFpd-TgIqNk9ninNWuk6XJTXMYkFSkpcdiTtEVL_uDbC8zeO3_-WufCFNJN3r7I8cF_7p4br_K7kt-FpVw/s16000/uqsay59_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.cee.ed.tum.de/bm/mitarbeiterinnen/felix-schneider/">Felix Schneider</a> (<a href="https://www.tum.de/">Technical University of Münich</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay59_slides_fschneider.pdf">slides</a>]</h4>
<br />
<h4 style="text-align: left; font-weight: bold;">Sparse Bayesian Learning for Rational Polynomial Chaos Expansion and Application in Structural Dynamics</h4>
<p>Surrogate models enable efficient propagation of uncertainties in computationally demanding models of physical systems. We employ surrogate models that draw upon polynomial bases to model the stochastic frequency response of structural dynamics systems subject to parameter uncertainty. Therein, we define a rational polynomial chaos expansion (rPCE) as the ratio of two polynomial chaos expansions. The rPCE is thereby specifically suitable for models that include poles in the system response. We apply least squares and Bayesian regression techniques to determine the numerator and denominator coefficients, which allows straightforward coupling of the proposed methods with existing black-box solvers.</p>
<p>This webinar focusses on a sparse Bayesian learning approach for the determination of the surrogate model coefficients in the rPCE. Due to the non-linearity of the surrogate with respsect to the denominator coefficients, we resort to a sequential solution strategy that utilizes the availability of a closed-form solution for the posterior distribution of the numerator coefficients. Subsequently, Laplace’s approximation is used to approximate the posterior distribution of the denominator coefficients. The coefficients and an optimal set of hyperparameters are then found in a sequential manner. We compare the performance with previously proposed strategies, which do not consider the uncertainty related to the denominator coefficients.</p>
<p>Joint work with Iason Papaioannou & Gerhard Müller</p>
<p>References:</p>
<ol>
<li>F. Schneider, I. Papaioannou, M. Ehre and D. Straub. <a href="https://dx.doi.org/10.1016/j.compstruc.2020.106223">Polynomial chaos based rational approximation in linear structural dynamics with parameter uncertainties</a>. Computers & Structures 233, 2020.</li>
<li>F. Schneider, I. Papaioannou, G. Müller, <a href="https://dx.doi.org/10.1002/nme.7182">Sparse Bayesian Learning for Complex‐Valued Rational Approximations</a>. International Journal for Numerical Methods in Engineering, 2022.</li>
<li>F. Schneider, I. Papaioannou, D. Straub, C. Winter, G. Müller, <a href="https://dx.doi.org/10.1016/j.ymssp.2021.108407">Bayesian parameter updating in linear structural dynamics with frequency transformed data using rational surrogate models</a>. Mechanical Systems and Signal Processing 166, 2022.</li>
</ol>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-34540828683355119762023-04-16T11:54:00.005-07:002023-05-12T02:28:06.384-07:00UQSay #58<div style="text-align: justify;">
<p>The fifty-eighth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, April 20, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay58_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhqDgEF1UKTOyG9EcfBINCnCGb_BWaYySEv-axzQ4kuOAojzssn63h_E5fCMBnEuG2Gmk6plKMwJjsNB-uwKBid6xa_xiTJVgUeW50kFBmXmMkG7Tceq8SmyURPVJtko4jmT3hlMWih9JQwzWonCkh0tPxSSzgy85sD2tBhSxCf3izj36KVMmGutvUQqA/s16000/uqsay58_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — Yingfan Wang (<a href="https://users.cs.duke.edu/~cynthia/lab.html">Interpretable ML Lab</a>, Duke Univ.) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay58_slides_ywang.pdf">slides</a>]</h4>
<br />
<h4 style="text-align: left; font-weight: bold;">Understanding principles for dimensionality reduction tools, and PaCMAP for data visualization</h4>
<p>Dimension reduction (DR) techniques such as t-SNE, UMAP, and TriMap have demonstrated impressive visualization performance on many real-world datasets. One tension that has always faced these methods is the trade-off between preservation of global structure and preservation of local structure. In this work, our main goal is to understand what aspects of DR methods are important for preserving both local and global structure. Towards the goal of local structure preservation, we provide several useful design principles for DR loss functions based on our new understanding of the mechanisms behind successful DR methods. Towards the goal of global structure preservation, our analysis illuminates that the choice of which components to preserve is important. We leverage these insights to design a new algorithm for DR, called Pairwise Controlled Manifold Approximation Projection (PaCMAP), which preserves both local and global structure.</p>
<p>Joint work with H. Huang, C. Rudin & Y. Shaposhnik.</p>
<p>References:</p>
<ol>
<li>Wang, Y., Huang, H., Rudin, C. & Shaposhnik, Y. (2021). Understanding how dimension reduction tools work: an empirical approach to deciphering t-SNE, UMAP, TriMAP, and PaCMAP for data visualization. Journal of Machine Learning Research 22.1, 9129-9201. <a href="https://dl.acm.org/doi/abs/10.5555/3546258.3546459">DOI:10.5555/3546258.3546459</a>,</li>
<li>Huang, H., Wang, Y., Rudin, C. & Browne, E.P. (2022). Towards a comprehensive evaluation of dimension reduction methods for transcriptomic data visualization. Communications biology 5, 719.<br /><a href="https://www.nature.com/articles/s42003-022-03628-x">DOI:10.1038/s42003-022-03628-x</a>.</li>
</ol>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-7496895135762111532023-03-24T13:19:00.004-07:002023-03-28T07:52:04.314-07:00UQSay #57<div style="text-align: justify;">
<p>The fifty-seventh UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, March 30, 2023.</p>
<p><b>WARNING: Unusual starting time: UQSay #57 will begin at <span style="color: red;">3 PM</span> (Paris time). Don't trust the poster.</b></p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay57_poster.pdf">
<img border="0" data-original-height="708" data-original-width="500" src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhSEayiA8GIHJCcLwGPqMCTTDKngFzJDhHC_VkGbPbuO8_CT5d0UJrmJztAQEM_yCgnwJDMYlut-DlKIksA9hGqrppAt-nS9BXsZ8DSALrL3c3Q88gNPb3Yu6pOMZ1RUen9tUmMV60d_Av3en_PPZsrOgvMWcaQ5MeTTysJqLllgqD_YBsdZ1xwGGWp1w/s16000/uqsay57_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>3–4 PM — <a href="https://cims.nyu.edu/~andrewgw/">Andrew Gordon Wilson</a> (New York University)</h4>
<br />
<h4 style="font-weight: bold; text-align: left;">Myths and Legends of Bayesian Deep Learning</h4>
<p>Bayesian inference makes more sense for modern neural networks than virtually every other model class, because these models can represent many compelling and complementary explanations for data, corresponding to different settings of their parameters. However, a number of myths have emerged about modern Bayesian deep learning. In this talk we will evaluate the following questions: (1) is Bayesian deep learning practical? (2) are standard (e.g. Gaussian) priors arbitrary and poor? (3) is "deep ensembles" a non-Bayesian competitor to standard approximate inference approaches? (4) does the common practice of posterior tempering, leading to "cold posteriors", mean the Bayesian posterior is poor? (5) is the marginal likelihood a reasonable way to select between trained networks? I will also discuss the success stories, future opportunities, and challenges in Bayesian deep learning.</p>
<p>References:</p>
<ol>
<li>Bayesian Deep Learning and a Probabilistic Perspective of Generalization (<a href="https://arxiv.org/abs/2002.08791">arXiv:2002.08791</a>)</li>
<li>What are Bayesian Neural Network Posteriors Really Like? (<a href="https://arxiv.org/abs/2104.14421">arXiv:2104.14421</a>)</li>
<li>Dangers of Bayesian Model Averaging under Covariate Shift (<a href="https://arxiv.org/abs/2106.11905">arXiv:2106.11905</a>)</li>
<li>On Uncertainty, Tempering, and Data Augmentation in Bayesian Classification (<a href="https://arxiv.org/abs/2203.16481">arXiv:2203.16481</a>)</li>
<li>Bayesian Model Selection, the Marginal Likelihood, and Generalization (<a href="https://arxiv.org/abs/2202.11678">arXiv:2202.11678</a>)</li>
<li>Residual Pathway Priors for Soft Equivariance Constraints (<a href="https://arxiv.org/abs/2112.01388">arXiv:2112.01388</a>)</li>
</ol>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-4456474803693210172023-03-10T07:45:00.003-08:002023-03-17T06:29:02.996-07:00UQSay #56<div style="text-align: justify;">
<p>The fifty-sixth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, March 16, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay56_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjSwhZpe3GFGvpC9XrMvC0V8VinvUDQo1Vl3Akq6RPSU9hmKjvXTEDgPQ82RUBWJVRrWHi41nQqtWLeLsWSF2gizYH8KlOAd82ys0s4H9bWTRe7jAlU_Lv4hpUGntolBxeOAI1r0nzfTbtKYZoh3wZJZKDF0kZzvsExaaZ4MSfEh-FwOnHm718Ht_vkwg/s16000/uqsay56_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://scholar.google.com/citations?user=uaJK95oAAAAJ&hl=fr">Paul Novello</a> (<a href="https://www.deel.ai/">DEEL</a> - <a href="https://www.irt-saintexupery.com/">IRT Saint-Exupery </a> - <a href="https://aniti.univ-toulouse.fr/">ANITI </a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay56_slides_pnovello.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Making Sense of Dependence: Efficient Black-box Explanations Using Dependence Measure</h4>
<p>In this talk, we present a new efficient black-box attribution method based on Hilbert-Schmidt Independence Criterion (HSIC), a dependence measure based on Reproducing Kernel Hilbert Spaces (RKHS). HSIC measures the dependence between regions of an input image and the output of a model based on kernel embeddings of distributions. It thus provides explanations enriched by RKHS representation capabilities. HSIC can be estimated very efficiently, significantly reducing the computational cost compared to other black-box attribution methods. Our experiments show that HSIC is up to 8 times faster than the previous best black-box attribution methods while being as faithful. Indeed, we improve or match the state-of-the-art of both black-box and white-box attribution methods for several fidelity metrics on Imagenet with various recent model architectures. Importantly, we show that these advances can be transposed to efficiently and faithfully explain object detection models such as YOLOv4. Finally, we extend the traditional attribution methods by proposing a new kernel enabling an ANOVA-like orthogonal decomposition of importance scores based on HSIC, allowing us to evaluate not only the importance of each image patch but also the importance of their pairwise interactions.</p>
<p>Joint work with T. Fel & D. Vigouroux.</p>
<p>References: <a href="https://doi.org/10.48550/arXiv.2206.06219">arXiv.2206.06219</a> & <a href="https://github.com/paulnovello/HSIC-Attribution-Method">github.com/paulnovello</a>. </p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-3750256417013179672023-02-13T23:02:00.006-08:002023-02-22T07:51:12.902-08:00UQSay #55<div style="text-align: justify;">
<p>The fifty-fifth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, February 16, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay55_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEh64VRxuyB3FcuE0jNNBrMwxUbyTqSdkVl3LO0qfFXyynIClsLGxUdPJmaCZNRr9UqmcE1B9E6aDrssyM33a9OEOyjZ8Yzn_VPl1F95YWwK9W4GhThMnegKblLsn-t2R2-J1j-XR_bgy80iNTG6ciCIicUoVV-b5wXFALudvNKj_3AGG6voGvV_kBPwYA/s708/uqsay55_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.cee.ed.tum.de/era/team/oindrila-kanjilal/">Oindrila Kanjilal</a> (<a href="https://www.cee.ed.tum.de/era/era-group/">T.U. Munich</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay55_slides_okanjilal.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Reliability assessment of structural dynamic systems by importance sampling</h4>
<p>Engineering structures are sometimes exposed to disastrous dynamic loading, such as strong wind and seismic motions. Structural failure due to these loads leads to significant economic loss and societal distress. Prediction of reliability during the service time is therefore essential in the design of new or integrity assessment of existing structures. Developments in computational mechanics provide tools to predict structural performance by means of simulation models. However, the parameters of these models, such as loading, material and geometric properties, deterioration processes and boundary conditions, can be seldom determined uniquely as they are affected by uncertainty and randomness. Model-based dynamic reliability assessment involves propagation of the input uncertainties through the model and exploration of the tails of the system response.</p>
<p>This webinar focuses on Monte Carlo simulation (MCS)-based computational approaches to estimate the reliability. The main challenge in applying MCS lies in controlling the sampling variance of the failure probability estimator; the aim is to obtain probability estimates of acceptable accuracy with a small number of computational model runs. In this talk we will discuss recently developed advanced Monte Carlo techniques based on important sampling to address this challenge. </p>
<p>Joint work with I. Papaioannou & D. Straub.</p>
<p>References:<ul>
<li>Kanjilal, O., Papaioannou, I., & Straub, D. (2021). Cross entropy-based importance sampling for first-passage probability estimation of randomly excited linear structures with parameter uncertainty. Structural Safety 91, 102090. <a href="https://doi.org/10.1016/j.strusafe.2021.102090">DOI:10.1016/j.strusafe.2021.102090</a>,</li>
<li>Kanjilal, O., Papaioannou, I., & Straub, D. (2022). Series system reliability of uncertain linear structures under Gaussian excitation by cross entropy-based importance sampling. ASCE Journal of Engineering Mechanics 148 (1), 04021136. <a href="https://doi.org/10.1061/(ASCE)EM.1943-7889.0002015">10.1061/(ASCE)EM.1943-7889.0002015</a>.</li>
</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-82760389909454312472023-01-27T22:45:00.002-08:002023-02-13T23:03:32.513-08:00UQSay #54<div style="text-align: justify;">
<p>The fifty-fourth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, February 2, 2023.</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay54_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjWfA8XS2llr-2bpxxFc1SKRqUfxWYXPt6Kq18ldnIPcRW1Qx3eIk56l-7cxyHSNs0DdLbOyCkrC3I2m4rXazlhNHP8oYhdYAUHt7XoZIoVel5Rc7KXsBGwmr08yozdfkYsyr85xsatKoadWYz8RoEvmXMd7QXxbBnz_1IXY475UZn1qlpQWCyvPDdo7w/s16000/uqsay54_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://scholar.google.fr/citations?user=61j2VawAAAAJ">Brian Staber</a> (<a href="https://www.safran-group.com/fr/groupe/innovation/safran-tech">Safran Tech</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay54_slides_bstaber.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Quantitative performance evaluation of Bayesian neural networks</h4>
<p>Due to the growing adoption of deep neural networks in many fields of science and engineering, modeling and estimating their uncertainties has become of primary importance. Various approaches have been investigated including Bayesian neural networks, ensembles, deterministic approximations, amongst others. Despite the growing litterature about uncertainty quantification in deep learning, the quality of the uncertainty estimates remains an open question. In this work, we attempt to assess the performance of several algorithms on sampling and regression tasks by evaluating the quality of the confidence regions and how well the generated samples are representative of the unknown target distribution. Towards this end, several sampling and regression tasks are considered, and the selected algorithms are compared in terms of coverage probabilities, kernelized Stein discrepancies, and maximum mean discrepancies.</p>
<p>Joint work with Sébastien Da Veiga (ENSAI).</p>
<p>Ref: <a href="https://arxiv.org/abs/2206.06779">arXiv:2206.06779</a></p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-84642655063724522552023-01-16T07:00:00.005-08:002023-02-13T23:03:38.726-08:00UQSay #53<div style="text-align: justify;">
<p>The fifty-third UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>January 19</b>, 2023.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay53_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEisfLZ7EfTQm24h0ztV9Tld4pFM7qYU0XjXhp9i7kt69jcmejHCjJTyX2fy_alXRcEcOER0svzbHxDFAybfO-JZn34YerSARZ80JoZX8iffd8DXJ5rGoRTiXC2d5xIHcxhDZKCpbwRCsBB7p5-pKLDZ3gFUbWSmTnimXnjTPK-lKCcDAt1sKmDxwbpBIA/s16000/uqsay53_poster_lowres.png" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="http://www.dim.uchile.cl/~ftobar/">Felipe Tobar</a> (<a href="https://idia.uchile.cl/">Initiative for Data & AI, Univ. de Chile</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay53_slides_ftobar.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Computationally-efficient initialisation of Gaussian processes: The generalised variogram method</h4>
<p>We present a computationally-efficient strategy to find the hyperparameters of a Gaussian process (GP) avoiding the computation of the likelihood function. Motivated by the fact that training a GP via ML is equivalent (on average) to minimising the KL-divergence between the true and learnt model, we set to explore different metrics/divergences among GPs that are computationally inexpensive and provide estimates close to those of ML. In particular, we identify the GP hyperparameters by projecting the empirical covariance or (Fourier) power spectrum onto a parametric family, thus proposing and studying various measures of discrepancy operating on the temporal or frequency domains. Our contribution extends the Variogram method developed by the geostatistics literature and, accordingly, it is referred to as the Generalised Variogram method (GVM). In this talk, we will start with a brief introduction to Gaussian processes, then present the proposed GVM and finally provide experimental validation using synthetic and real-world data.</p>
<p>Joint work with Elsa Cazelles & Taco de Wolff.</p>
<p>Ref: <a href="https://arxiv.org/abs/2210.05394">arXiv:2210.05394</a>.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-26210950847318839672023-01-02T00:24:00.000-08:002023-01-02T00:24:32.826-08:00UQSay #52<div style="text-align: justify;">
<p>The fifty-second UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>January 5</b>, 2023.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay52_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgB7lgoHPl4_8O2q413M4wi5tZINieGhuL-4m2VOYtrd72qbkegDccd-16TXf6gCQGtJYyFHDnP13h7Kg_UnyjO8S1Nf09xE4XVCj3vkfHi3eMo0x3eOEwTvly6BurHsPD39ZDal73HyULHfEZldmr9TpGZIRBLo_AtS9uHTebedamuiMgkxnqgEqrNRg/s16000/uqsay52_poster_lowres.jpg" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.maths.dur.ac.uk/users/georgios.karagiannis/">Georgios Karagiannis</a> (<a href="https://www.durham.ac.uk/departments/academic/mathematical-sciences/">Durham University</a>)</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Bayesian spanning treed co-kriging for high dimensional output emulation</h4>
<p>We propose a new Bayesian emulator, called Bayesian spanning treed co-kriging, suitable to analyze computer models with non-stationary massive outputs in the multifidelity setting. Our motivation comes from a real-life application with a storm surge simulator. Given certain assumptions on the Bayesian model, we introduce a suitable stochastic mechanism that facilitates predictions in a principal manner. The good performance of our method is demonstrated in benchmark examples, while our method is implemented for the analysis of a surge simulator given simulations at different fidelity levels.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-16119871911823518492022-11-12T12:43:00.008-08:002022-11-18T02:05:23.494-08:00UQSay #51<div style="text-align: justify;">
<p>The fifty-first UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>November 17</b>, 2022.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay51_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjo4Mnh5ScmDhZk6C20JeX3idF4d8iSsWydLsd8BHP9RV_jtTNjngPmnM8ewaggjtSyi1lgUn-h9mvutqzZLmAYra5jrQ_g-yHxYcgD2xuSrDtzqPxDT4mNVO1RPrV4e6Z_0my2aZpP_BeHvKdNM54TKMhYyoqLAZRbQmOC5H_k4YEEkssm2RjGiF2DSQ/s16000/uqsay51_poster_lowres.jpg" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="http://math.univ-lyon1.fr/~mercadier/">Cécile Mercadier</a> (<a href="http://math.univ-lyon1.fr/">Institut Camille Jordan</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay51_slides_cmercadier.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Hoeffding–Sobol and Möbius decompositions for (tail-)dependence analysis</h4>
<p>Methods to analyse dependence and tail dependence are well established. Using for instance the copula function or the stable tail dependence function, and their empirical versions, one can construct non parametric statistics, parametric inference, as well as testing or resampling procedures. My talk will reflect upon the use of g sensitivity analysis for extreme value theory and copula modeling. Through my recent publications, I will explain what their links are and the benefit in mixing these domains.</p>
<p>Joint work with Christian Genest, Paul Ressel & Olivier Roustant.</p>
<p>Refs:</p>
<ul>
<li> C. Mercadier, O. Roustant & C. Genest (2022). Linking the Hoeffding–Sobol and Möbius formulas through a decomposition of Kuo, Sloan, Wasilkowski, and Wozniakowski. <a href="https://doi.org/10.1016/j.spl.2022.109419">Statistics & Probability Letters</a>, vol. 185 [<a href="https://hal.archives-ouvertes.fr/hal-03220809/">hal-03220809</a>],</li>
<li>C. Mercadier & P. Ressel (2021). Hoeffding–Sobol decomposition of homogeneous co-survival functions: from Choquet representation to extreme value theory application. <a href="https://doi.org/10.1515/demo-2021-0108">Dependence Modeling</a>, 9(1):179–198 [<a href="https://hal.archives-ouvertes.fr/hal-03200817/">hal-03200817</a>],</li>
<li>C. Mercadier & O. Roustant (2019). The tail dependograph. <a href="https://doi.org/10.1007/s10687-019-00345-3">Extremes</a>, 22:343–372 [<a href="https://hal.archives-ouvertes.fr/hal-01649596">hal-01649596</a>].</li>
</ul>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-76074308046616625022022-10-04T08:48:00.002-07:002022-11-15T05:48:48.776-08:00UQSay #50<div style="text-align: justify;">
<p>The fiftieth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>October 13</b>, 2022.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay50_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhx6YLPZH2T2Rrpo2E3woSDvR7uW0PL0mrZtnirhg-ig90j9cadrAF7YtC0A1UZsw_13tuJV2msa7P8fWe_4860c2dwYlvBh5TfK5QI5QRxPEhxP2HM8Oaosv_dcEt1t-owiF_QXqLhnx-JNcpABURG-65uVv915OofQqR1uKsKn-u3dpTQbyuXma_G0w/s16000/uqsay50_poster_lowres.jpg" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.imo.universite-paris-saclay.fr/fr/perso/gilles-stoltz/">Gilles Stoltz</a> (<a href="https://www.imo.universite-paris-saclay.fr/">LMO, Université Paris-Saclay</a> - <a href="https://www.cnrs.fr/">CNRS</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay50_slides_gstoltz.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Multi-armed bandit problems: a statistical view, focused on lower bounds</h4>
<p>Multi-armed bandit problems correspond to facing K unknown probability distributions, having to sequentially pull one of them, and observing a realization thereof at each pull. Two goals will be considered.</p>
<p>(1) The realizations are payoffs, and the sum of these payoffs is to be maximized. This goal is achieving by minimizing regret, which is defined as the expected performance of the best arm minus the expected sum of payoffs achieved by a strategy. Two types of bounds may be defined, depending on whether they may depend on the specific bandit problem or only on the model (the class of possible distributions). We will recall classical strategies like UCB and MOSS, as well as a new strategy combining both, called KL-UCB-Switch. We will review upper bounds on the regret and detail which lower bounds may be achieved, and how. We will deal with one interesting extension, the adaptation to the unknown range of the distributions, i.e., when the distribution are supported on a compact interval that is unknown as well.</p>
<p>The case of regret minimization is very well understood in the literature, contrary to:</p>
<p>(2) A second goal can be to identify the best arm, i.e., control the probability that after T observations (sampled adaptively) the strategy does not identify the arm with the highest expectation. This is called best arm identification with a fixed budget. Limited results are available. We will describe a typical strategy, called successive rejects, that drops one distribution after the other after horse racing them. We will also indicate how we are currently laying the foundations of a non-parametric approach to this problem, based on KL divergences, as opposed to typical approaches based on differences between expectations.</p>
<p>Joint work with Antoine Barrier, Aurélien Garivier, Hédi Hadiji & Pierre Ménard.</p>
<p>Refs:</p>
<ul>
<li>KL-UCB-Switch: <a href="https://jmlr.org/papers/volume23/20-717/20-717.pdf">JMLR, 23(179):1−66, 2022</a>,</li>
<li>Lower bounds for regret minimization: <a href="https://pubsonline.informs.org/doi/10.1287/moor.2017.0928">MOOR, 4(2):377-766, 2019</a> [<a href="https://hal.archives-ouvertes.fr/hal-01785705v2">HAL</a>],</li>
<li>Adaptation to the unknown range: <a href="https://hal.archives-ouvertes.fr/hal-02794382/">HAL-02794382</a>,</li>
<li>Best-arm identification: <a href="https://hal.archives-ouvertes.fr/hal-03792668/">HAL-03792668</a>.</li>
</ul>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-27852654492533245852022-09-21T12:25:00.004-07:002022-10-07T03:10:10.603-07:00UQSay #49<div style="text-align: justify;">
<p>The forty-ninth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>September 29</b>, 2022.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay49_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgIDoKC20s9PswOi8-XCYH4l9QEK9WYczgiZVFgpC-MNKzeVRcPAlga3eN1i42SawHqXvnxW_f22fQNMZGCvWKcyQzg4-KFOCPzitoIaFlb2QS39vTHkePm7YZY_xvC3PD0_B18HgdY2ImMIWK8oQ9Yv3qYM7peQVWScEkhhmMSRF1802Ebelb50ympfA/s16000/uqsay49_poster_lowres.jpg" />
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://latzplacian.org/">Jonas Latz</a> (<a href=" https://www.hw.ac.uk/uk/schools/mathematical-computer-sciences.htm">Heriot-Watt University, Edinburgh</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay49_slides_jlatz.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Stochastic gradient descent in continuous time: discrete and continuous data</h4>
<p>Optimisation problems with discrete and continuous data appear in statistical estimation, machine learning, functional data science, robust optimal control, and variational inference. The "full" target function in such an optimisation problem is given by the integral over a family of parameterised target functions with respect to a discrete or continuous probability measure. Such problems can often be solved by stochastic optimisation methods: performing optimisation steps with respect to the parameterised target function with randomly switched parameter values. In this talk, we discuss a continuous-time variant of the stochastic gradient descent algorithm. This so-called stochastic gradient process couples a gradient flow minimising a parameterised target function and a continuous-time 'index' process which determines the parameter.</p>
<p>We first briefly introduce the stochastic gradient processes for finite, discrete data which uses pure jump index processes. Then, we move on to continuous data. Here, we allow for very general index processes: reflected diffusions, pure jump processes, as well as other Lévy processes on compact spaces. Thus, we study multiple sampling patterns for the continuous data space. We show that the stochastic gradient process can approximate the gradient flow minimising the full target function at any accuracy. Moreover, we give convexity assumptions under which the stochastic gradient process with constant learning rate is geometrically ergodic. In the same setting, we also obtain ergodicity and convergence to the minimiser of the full target function when the learning rate decreases over time sufficiently slowly.</p>
<p>Joint work with Kexin Jin, Chenguang Liu & Carola-Bibiane Schönlieb.</p>
<p>Refs: <a href="https://link.springer.com/article/10.1007/s11222-021-10016-8">DOI:10.1007/s11222-021-10016-8</a>, <a href="https://arxiv.org/abs/2112.03754">arXiv:2112.03754</a>, <a href="https://arxiv.org/abs/2203.11555">arXiv:2203.11555</a>.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinators</b>: Julien Bect (L2S) & Sidonie Lefebvre (ONERA)</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-14589348368151126542022-05-24T03:01:00.003-07:002022-06-10T06:33:45.022-07:00UQSay #48<div style="text-align: justify;">
<p>The forty-eighth UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>June 2</b>, 2022.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay48_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEj8zd9YBcKdVX8ob_mZr0rOTpBZUDVJ16Rh6pBH0hN1haAMBlHHR1yUS9eP-LiQhtpEa53MktHN6ZZbRbWSuHGEOpMHMNznidV4dzu-JHs8TL4aY5VogsoFzCmOHpQLmYNme6QgXimpdCuKQODppMiMD99bwMUCVrO5SNshLonvGIs-AvZIq7GGjw2sYQ/s16000/uqsay48_poster_lowres.jpg" /></a></div><br /><a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay48_poster.pdf"
/>
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://www.valentinresseguier.com/">Valentin Resseguier</a> (<a href="https://www.scalian.com/en/expertise/rd-innovation/">Scalian Innovation Lab</a>, <a href="https://www6.rennes.inrae.fr/opaale/RECHERCHE/Equipes-de-recherche/ACTA">INRAE</a>) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay48_slides_vresseguier.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Fast generation of prior for Bayesian estimation problems in fluid mechanics</h4>
<p>We are interested in real-time estimation and short-term forecasting of 3D fluid flows, using limited computational resources. This is possible through the coupling between data, numerical simulations and sparse fluid flow measurements. Here, the term data refers to numerical simulation outputs. To achieve these ambitious goals, synthetic (i.e. simulated) data and intrusive surrogate models drastically reduce the problem dimensionality – typically from 10 7 to 10. Unfortunately, even with corrections, the accumulated errors of these surrogate models increase rapidly over time due to the chaotic and intermittent nature of fluid mechanics. Therefore, deterministic predictions are hardly possible outside the learning time interval. Data assimilation can alleviate these problems by (i) providing a set of simulations covering probable futures (without increasing the computational cost) and (ii) constraining these online simulations with measurements.</p>
<p>We addressed this Uncertainty Quantification (UQ) problem (i) with a multi-scale physically-based stochastic parameterization called "Location uncertainty models" (LUM) [1-3] and new statistical estimators based on stochastic calculus, signal processing and physics [3]. The deterministic ROM coefficients are obtained by a Galerkin projection whereas the correlations of the noises are estimated from the residual velocity, the physical model structure, and the evolution of the resolved modes. We solved problem (ii) with a particle filter [4].</p>
<p>Whether we consider UQ [3] or DA [4] applications, our method greatly exceeds the state of the art, for ROM degrees of freedom smaller than 10 and moderately turbulent 3D flows (Reynolds number up to 300).</p>
<p>Joint work A. M. Picard & M. Ladvig (Scalian), and D. Heitz (INRAE).</p>
<p>Refs: [1] <a href="https://hal.archives-ouvertes.fr/hal-01391420">hal-01391420</a>, [2] <a href="https://hal.archives-ouvertes.fr/hal-02558016">hal-02558016</a>, [3] <a href="https://hal.archives-ouvertes.fr/hal-03169957">hal-03169957</a> & [4] <a href="https://hal.archives-ouvertes.fr/hal-03445455">hal-03445455</a>.</p>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinator</b>: Julien Bect (L2S).
</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.comtag:blogger.com,1999:blog-3196435405422718870.post-25261660447964867242022-05-13T00:36:00.003-07:002022-06-01T13:29:17.586-07:00UQSay #47<div style="text-align: justify;">
<p>The forty-seventh UQSay seminar on <a href="https://www.uqsay.org/scope">UQ, DACE and related topics</a> will take place <b>online</b> on Thursday afternoon, <b>May 19</b>, 2022.
</p>
<div style="margin-top: 25pt; text-align: center;">
<a href="https://www.l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay47_poster.pdf">
<img border="0"
data-original-height="708"
data-original-width="500"
src="https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEj1k13d-2KJ1dEzKc9n21XmPf7C5OG5CkKklJHz00Dep2gezUp2Q6IaA-5BlodXEHkin7ESKe-I5wkuKLTIiD7BKHIEDv6q43c3h80k0228BeEaMJaEUsIdi9G1ov6gRA5K4P8o2IAQEhnNEEjmEI1-yOfDO5qh-ydVOZFVpLHI5Hfjrn8G3IbWwNacBw/s16000/uqsay47_poster_lowres.jpg"
/>
</a>
</div>
<div style="margin-top: 25pt;">
<h4>2–3 PM — <a href="https://scholar.google.fr/citations?user=PyB6SQwAAAAJ">Mélanie Rochoux</a> (<a href="https://cerfacs.fr/umr-5318-climat-environnement-couplages-et-incertitudes-ceci/">CECI</a>, Cerfacs, CNRS) — [<a href="https://l2s.centralesupelec.fr/wp-content/uploads/uqsay/uqsay47_slides_mrochoux.pdf">slides</a>]</h4>
<br/>
<h4 style="text-align: left; font-weight: bold;">Assimilating fire front position and emulating boundary-layer flow simulations for wildland fire behavior ensemble prediction and reanalysis</h4>
<p>Monitoring wildfire behavior has recently emerged as a key public policy issue due to the occurrence of extreme events, in particular in the Euro-Mediterranean area that is exposed to more frequent and more severe wildfires under climate change. Key to this modeling is the development of an event-scale numerical simulation capability as a means to understand and predict the interactions between the atmosphere and the wildfire that drive its behavior.</p>
<p>In this framework, my research aims at designing and evaluating a wildland fire behavior reanalysis capability to reconstruct as best as possible wildland fire progression at landscape-to-atmospheric scales. This approach combines information coming from a coupled atmosphere/fire model (Costes et al. 2021) and from airborne thermal infrared images (Paugam et al. 2021) through an ensemble-based data assimilation algorithm that infers more realistic environmental factors and estimates the time-evolving fire front position.</p>
<p>My talk will provide an overview of the different components required to build this reanalysis capability, with two main focus: i) a front data assimilation methodology to address position errors in the fire front progression (Rochoux et al. 2018; Zhang et al. 2019), and ii) a non-intrusive reduced-order modeling approach combining principal component analysis and adaptive Gaussian processes to accurately and efficiently explore the physical parameter space and predict the atmospheric boundary-layer flow patterns (Nony et al. 2021). In the long-term, these methods will be applied to the Meso-NH/Blaze coupled atmosphere/fire model to design a wildland fire behavior ensemble prediction and reanalysis capability.</p>
<p>Joint work with Bastien Nony & Thomas Jaravel (Cerfacs), Didier Lucor (LISN), Annabelle Collin & Philippe Moireau (Inria), Cong Zhang & Arnaud Trouvé (University of Maryland).</p>
<p>Refs:</p>
<ul>
<li>M.C. Rochoux, A. Collin, C. Zhang, A. Trouvé, D. Lucor and P. Moireau (2018). Front shape similarity measure for shape-oriented sensitivity analysis and data assimilation for eikonal equation. ESAIM: Proceedings and Surveys, EDP Sciences, 63:258–279, DOI:<a href="https://doi.org/10.1051/proc/201863258">10.1051/proc/201863258</a>.</li>
<li>C. Zhang, A. Collin, P. Moireau, A. Trouvé and M.C. Rochoux (2019). State-parameter estimation approach for data-driven wildland fire spread modeling: application to the 2012 RxCADRE S5 field-scale experiment. Fire Safety Journal, 105:286–299, DOI:<a href="https://doi.org/10.1016/j.firesaf.2019.03.009"><10.1016/j.firesaf.2019.03.009</a>.</li>
<li>B.X. Nony, M.C. Rochoux, D. Lucor and T. Jaravel (2021). Compound parametric metamodeling of large-eddy simulations for micro-scale atmospheric dispersion. 20th International Conference on Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes, Tartu (Estonia), 14–18 June, 2021.</li>
<li>A. Costes, M.C. Rochoux, C. Lac and V. Masson (2021) Subgrid-scale fire front reconstruction for ensemble coupled atmosphere-fire simulations of the FireFlux I experiment. Fire Safety Journal, 126:103475, DOI:<a href="https://doi.org/10.1016/j.firesaf.2021.103475">10.1016/j.firesaf.2021.103475</a>.</li>
<li>R. Paugam, M.J. Wooster, W.E. Mell, M.C. Rochoux, J-B. Filippi, G. Rücker, O. Frauenberger, E. Lorenz, W. Schroeder and N. Govendor (2021). Orthorectification of helicopter-borne high resolution experimental burn observation from infra red handheld imagers. Remote Sensing, 13(23):4913, DOI:<a href="https://doi.org/10.3390/rs13234913">10.3390/rs13234913</a>.</li>
</ul>
</div>
<div style="margin-top: 25pt;">
<p><b>Organizing committee</b>: Pierre Barbillon (MIA-Paris), Julien Bect (L2S), Nicolas Bousquet (EDF R&D), Amélie Fau (LMPS), Filippo Gatti (LMPS), Bertrand Iooss (EDF R&D), Alexandre Janon (LMO), Sidonie Lefebvre (ONERA), Didier Lucor (LISN), Emmanuel Vazquez (L2S).</p>
<p><b>Coordinator</b>: Julien Bect (L2S).
</p>
</div>
<div style="margin-top: 25pt;">
<p>
<b>Practical details</b>: the seminar will be held online using Microsoft Teams.
</p>
<p>
If you want to attend this seminar (or any of the forthcoming online UQSay seminars), and <b>if you do not already have access to the UQSay group on Teams</b>, simply <a href="mailto:julien.bect@centralesupelec.fr?subject=UQSay%20Teams&body=Please%20invite%20me%20to%20UQSay%21" target="_blank">send an email</a> and you will be invited. Please specify which email address the invitation must be sent to (this has to be the address associated with your Teams account).
</p>
<p>
You will find the link to the seminar on the "General" UQSay channel on Teams, approximately 15 minutes before the beginning.
</p>
<p style="font-style: italic;">
The technical side of things: you can use Teams either directly from your web browser or using the "fat client", which is available for most platforms (<a href="https://www.microsoft.com/en-us/microsoft-365/microsoft-teams/download-app">Windows, Linux, Mac, Android & iOS</a>). We strongly recommend the latter option whenever possible. Please give it a try before the seminar to anticipate potential problems.
</p>
</div>
</div>
Julien Becthttp://www.blogger.com/profile/16442227081590376133noreply@blogger.com