Program modification

Author: CMoucer

february_2023/program.html

@@ -230,15 +230,17 @@ <h2>Monday the 13th</h2>
                 </details>
                 <br>
                 <details>
-                <summary><i>Manu Upadhyaya (Lund University)</i> - <strong>Tight Lyapunov function existence analysis for first-order
-                    methods  </strong> </summary>
-                    We present a unifying framework for establishing linear convergence rates for first-order methods used
-                    to solve convex optimization problems. To accomplish this, we derive a necessary and sufficient
-                    condition for verifying the existence of a quadratic Lyapunov function for the algorithm and problem
-                    class under consideration. This allows us to produce tight convergence certificates for the setting
-                    at hand (i.e., producing worst-case certificates and matching worst-case examples).
+                <summary><i>Yassine Kamri (UCLouvain)</i> - <strong>On the worst-case analysis of block coordinate-wise algorithms  </strong> </summary>
+                Block coordinate-wise algorithms are an essential class of first-order optimization methods widely used
+                    to solve large-scale optimization problems. However, their worst-case performance is still not well
+                    understood and their practical success has not yet been explained by existing convergence analyses.
+                    We analyze the worst-case behavior of cyclic versions of Block coordinate-wise algorithms in the
+                    context of unconstrained optimization of convex functions with coordinate-wise Lipschitz gradients.
+                    For this purpose, we rely on the recently proposed Performance Estimation Problem (PEP) and develop a
+                    new characterization for this class of function, from which we obtain necessary interpolation conditions.
                     <br>
-                    This is joint work with Sebastian Banert, Adrien Taylor and Pontus Giselsson."
+                    In this paper, we present a unifying framework for the worst-case analysis of Block coordinate-wise
+                    algorithms, and in some situations, we are able to substantially outperform the best current bounds.
                 </details>
             </td>
         </tr>
@@ -361,28 +363,14 @@ <h2>Tuesday the 14th</h2>
                     in the convex case are a particular case of our analysis. Finally, we identify the optimal constant
                     step size that minimizes the worst-case of the gradient method applied to hypoconvex functions.
                     </details>
-                <br>
-                <details>
-                <summary><i>Yassine Kamri (UCLouvain)</i> - <strong>On the worst-case analysis of block coordinate-wise algorithms  </strong> </summary>
-                Block coordinate-wise algorithms are an essential class of first-order optimization methods widely used
-                    to solve large-scale optimization problems. However, their worst-case performance is still not well
-                    understood and their practical success has not yet been explained by existing convergence analyses.
-                    We analyze the worst-case behavior of cyclic versions of Block coordinate-wise algorithms in the
-                    context of unconstrained optimization of convex functions with coordinate-wise Lipschitz gradients.
-                    For this purpose, we rely on the recently proposed Performance Estimation Problem (PEP) and develop a
-                    new characterization for this class of function, from which we obtain necessary interpolation conditions.
-                    <br>
-                    In this paper, we present a unifying framework for the worst-case analysis of Block coordinate-wise
-                    algorithms, and in some situations, we are able to substantially outperform the best current bounds.
-                </details>
             </td>
          </tr>
     </table>
 
   <table id="break 3">
         <tr>
             <td class="date" rowspan="3">
-                10:15am
+                9:50am
             </td>
             <td class="title">
                 Break
@@ -393,7 +381,7 @@ <h2>Tuesday the 14th</h2>
     <table id="Shuvo">
         <tr>
             <td class="date" rowspan="3">
-                10:30am
+                10:10am
             </td>
             <td class="title">
                 Long Talk - Shuvomoy Das Gupta (MIT)
@@ -440,7 +428,7 @@ <h2>Tuesday the 14th</h2>
   <table id="break 4">
         <tr>
             <td class="date" rowspan="3">
-                11:30am
+                11:10am
             </td>
             <td class="title">
                 Break
@@ -451,7 +439,7 @@ <h2>Tuesday the 14th</h2>
   <table id="Talks Session 4">
         <tr>
             <td class="date" rowspan="3">
-                11:45am
+                11:30am
             </td>
             <td class="title">
                 Talks Session 4
@@ -496,7 +484,7 @@ <h2>Tuesday the 14th</h2>
     <table id="lunch 2">
         <tr>
             <td class="date" rowspan="3">
-                1:00pm
+                12:45pm
             </td>
             <td class="title">
                 Lunch
@@ -511,11 +499,12 @@ <h2>Tuesday the 14th</h2>
                 2:00pm
             </td>
             <td class="title">
-                Long Talk - Pontus Gisselson (Tilburg Uniersity) <addinfo> - part of <a href="https://uclouvain.be/en/research-institutes/icteam/inma/seminars.html" target="_blank">INMA Seminar series</a>  </addinfo>
+                Long Talk - Pontus Gisselson (Lund University) <addinfo> - part of <a href="https://uclouvain.be/en/research-institutes/icteam/inma/seminars.html" target="_blank">INMA Seminar series</a>  </addinfo>
             </td>
         </tr>
         <tr>
             <td class="speaker">
+
                 <details>
                 <summary>Biography</summary>
                 Pontus Giselsson is an Associate Professor at the Department of Automatic Control at Lund University,
@@ -527,6 +516,8 @@ <h2>Tuesday the 14th</h2>
         </tr>
         <tr>
             <td class="abstract">
+                This is a joint work with Manu Updhyaya, Sebastian Banert and Adrien Taylor.
+                <br>
                 <strong>Tight Lyapunov function existence analysis for first-order methods </strong>
                 <details>
                 <summary>Abstract</summary>