<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class="">
<div class=""><b class="">Ruben Wiersma</b></div>
<div class=""><b class="">PhD Student</b></div>
<div class=""><b class="">University of Technology Delft,</b> <b class="">Netherlands<br class="">
</b></div>
<div class=""><br class="">
</div>
<div class=""><b class="">Wednesday, April 6th</b><br class="">
</div>
<div class=""><b class="">11 AM - 12:30 PM</b><br class="">
<b class="">Remote</b>: <a href="https://urldefense.com/v3/__https://tau-ac-il.zoom.us/j/89222884685__;!!BpyFHLRN4TMTrA!oE5LUAJf-C90WRH38BjpAZZPZ4orBKo4Lk89HRA44yJSh4lWHgqaW3n4WK-PoJnwEQ$" target="_blank" class="">Zoom</a><br class="">
<b class="">In Person</b>: <b class="">John Crerar Library, Room 346</b><br class="">
<div class=""><b class="">DeltaConv: Anisotropic Geometric Deep Learning with Exterior Calculus</b></div>
<br class="">
Learning from 3D data with convolutional neural networks (CNNs) has rapidly gained momentum, motivated by the success of CNNs on images and the increased availability of 3D data. In this talk, we discuss how to construct anisotropic convolutions that work directly
on surfaces. This is challenging because of the lack of a global coordinate system for tangential directions on surfaces. We'll take an in-depth look at DeltaConv, a convolution operator for point clouds that builds on well known geometric operators from vector
calculus. Because these operators are defined on scalar- and vector-fields, the convolutions separate into a scalar- and a vector-stream. The vector stream enables the network to explicitly represent, evaluate, and process directional information. DeltaConv
is robust and simple to implement and shows improved accuracy compared to state-of-the-art approaches on several benchmarks, while also speeding up training and inference.<br class="">
<br class="">
<b class="">Bio</b>: Ruben Wiersma is a third-year PhD student at the Computer Graphics and Visualization group at the University of Technology Delft. In his research he seeks to connect computer graphics- and geometry processing techniques to applications
in machine learning and painting analysis. Ruben is advised by Dr. Klaus Hildebrandt and supervised by Prof. Elmar Eisemann and Prof. Joris Dik. Outside of research, Ruben enjoys playing the piano, biking, and walking in the (flat) outdoors of the Netherlands.<br class="">
</div>
<div class=""><br class="">
</div>
<div class=""><b class="">Host: </b>Rana Hanocka</div>
<div class=""><br class="">
</div>
<div class=""><br class="">
</div>
<div class="">
<div>Holly Santos<br class="">
Executive Assistant to Michael J. Franklin, Chairman<br class="">
Department of Computer Science<br class="">
The University of Chicago<br class="">
5730 S Ellis Ave-217 Chicago, IL 60637<br class="">
P: 773-834-8977<br class="">
<a href="mailto:hsantos@uchicago.edu" class="">hsantos@uchicago.edu</a></div>
<div class=""><br class="">
</div>
<br class="Apple-interchange-newline">
</div>
<br class="">
</body>
</html>