Frederik Wenkel

Frederik Wenkel serves as a PhD candidate in Applied Mathematics at Université de Montréal and Mila - Quebec AI Institute, where he specializes in geometric deep learning and graph neural networks. His research focuses on developing innovative approaches to overcome limitations in graph convolutional networks, particularly addressing the oversmoothness problem in graph-based deep learning. Building on his strong mathematical foundation from Technical University of Munich, where he completed both his bachelor's and master's degrees, he explores applications of these techniques across diverse domains including social networks, biochemistry, and finance. His work combines geometric scattering transforms with residual convolutions to improve node discrimination while maintaining structural awareness in graph data. Through his research, he contributes to advancing the field of geometric deep learning, particularly in addressing challenges where traditional deep learning methods face limitations with non-Euclidean data structures.