Create publication page for GEPS

_publications/2024_geps.md

@@ -0,0 +1,76 @@
+---
+layout: publication
+title: "GEPS: Boosting Generalization in Parametric PDE Neural Solvers through Adaptive Conditioning" 
+image: 
+hide: true
+category: [generalization, dynamics, prediction]
+authors: Armand Kassaï Koupaï, Jorge Mifsut-Benet, Yuan Yin, Jean-Noël Vittaut, Patrick Gallinari
+venue: NeurIPS
+venue_long: Conference on Neural Information Processing Systems
+year: 2024
+month: 12
+code_url: https://github.com/itsakk/geps
+paper_url: https://arxiv.org/abs/2410.23889
+blog_url: https://geps-project.github.io
+slides_url: 
+bib_url: 
+permalink: /publications/geps/
+---
+
+<h1 align="center"> {{page.title}} </h1>
+<!-- Simple call of authors -->
+<!-- <h3 align="center"> {{page.authors}} </h3> -->
+<!-- Alternatively you can add links to author pages -->
+<h3 align="center"> <a href="https://itsakk.github.io/">Armand Kassaï Koupaï</a> &nbsp;&nbsp; <a href="https://www.isir.upmc.fr/personnel/mifsutbenet/">Jorge Mifsut-Benet</a> &nbsp;&nbsp; <a href="https://yuan-yin.github.io">Yuan Yin</a> &nbsp;&nbsp; <a href="https://webia.lip6.fr/~vittaut/">Jean-Noël Vittaut</a> &nbsp;&nbsp; <a href="https://pages.isir.upmc.fr/gallinari/">Patrick Gallinari</a> 
+
+
+<h3 align="center"> {{page.venue}} {{page.year}} </h3>
+
+<div align="center">
+  <p>
+    {% if page.paper_url %}
+    <a href="{{ page.paper_url }}"><i class="far fa-file-pdf"></i> Paper</a>&nbsp;&nbsp;
+    {% endif %}
+    {% if page.code_url %}
+    <a href="{{ page.code_url }}"><i class="fab fa-github"></i> Code</a> &nbsp;&nbsp;
+    {% endif %}
+    {% if page.blog_url %}
+    <a href="{{ page.blog_url }}"><i class="fab fa-blogger"></i> Blog</a> &nbsp;&nbsp;
+    {% endif %}
+    {% if page.slides_url %}
+    <a href="{{ page.slides_url }}"><i class="far fa-file-pdf"></i> Slides</a>&nbsp;&nbsp;
+    {% endif %}
+    {% if page.bib_url %}
+    <a href="{{ page.bib_url}}"><i class="far fa-file-alt"></i> BibTeX</a>&nbsp;&nbsp;
+    {% endif %}
+  </p>
+</div>
+
+
+<div class="publication-teaser">
+    <img src="../../{{ page.image }}" alt="project teaser"/>
+</div>
+
+
+<hr>
+
+<h2  align="center"> Abstract</h2>
+
+<p align="justify">Solving parametric partial differential equations (PDEs) presents significant challenges for data-driven methods due to the sensitivity of spatio-temporal dynamics to variations in PDE parameters. Machine learning approaches often struggle to capture this variability. To address this, data-driven approaches learn parametric PDEs by sampling a very large variety of trajectories with varying PDE parameters. We first show that incorporating conditioning mechanisms for learning parametric PDEs is essential and that among them, \textit{adaptive conditioning}, allows stronger generalization. As existing adaptive conditioning methods do not scale well with respect to the number of PDE parameters, we propose GEPS, a simple adaptation mechanism to boost GEneralization in Pde Solvers via a first-order optimization and low-rank rapid adaptation of a small set of context parameters. We demonstrate the versatility of our approach for both fully data-driven and for physics-aware neural solvers. Validation performed on a whole range of spatio-temporal forecasting problems demonstrates excellent performance for generalizing to unseen conditions including initial conditions, PDE coefficients, forcing terms and solution domain.</p>
+
+<hr>
+<hr>
+
+<h2  align="center">BibTeX</h2>
+<left>
+  <pre class="bibtex-box">
+@inproceedings{kassai2024geps,
+  title={GEPS: Boosting Generalization in Parametric PDE Neural Solvers through Adaptive Conditioning},
+  author={Kassaï Koupaï, Armand and Mifsut Benet, Jorge and Vittaut, Jean-Noël and Gallinari, Patrick},
+  booktitle={38th Conference on Neural Information Processing Systems (NeurIPS 2024)},
+  year={2024}
+}
+</pre>
+</left>
+
+<br>