BIU learning club – Moshe Eliasof – Improving Graph Neural Networks with Learnable Propagation Operators

Location:

Engineering building (1103), room 329

Title:

Improving Graph Neural Networks with Learnable Propagation Operators

Abstract:

Graph Neural Networks (GNNs) are limited in their propagation operators. In many cases, these operators often contain non-negative elements only and are shared across channels, limiting the expressiveness of GNNs. Moreover, some GNNs suffer from over-smoothing, limiting their depth. On the other hand, Convolutional Neural Networks (CNNs) can learn diverse propagation filters, and phenomena like over-smoothing are typically not apparent in CNNs. In this work, we bridge these gaps by incorporating trainable channel-wise weighting factors ω to learn and mix multiple smoothing and sharpening propagation operators at each layer. Our generic method is called wGNN, and is easy to implement. We study two variants: wGCN and wGAT. For wGCN, we theoretically analyze its behavior and the impact of w on the obtained node features. Our experiments confirm these findings, demonstrating and explaining how both variants do not over-smooth. Additionally, we experiment with 15 real-world datasets on node- and graph-classification tasks, where our wGCN and wGAT perform on par with state-of-the-art methods.

Short Bio:

Moshe Eliasof is a Ph.D. student at the Computer Science department at Ben-Gurion University of the Negev, advised by Prof. Eran Treister. In October 2023, he will join the CIA group at Cambridge University as a postdoctoral researcher, working with Prof. Carola-Bibiane Schönlieb. Moshe’s research is focused on Graph Neural Networks (GNNs), mostly on the interpretation of GNNs as dynamical systems. Moshe earned his B.Sc. in Computer Science at the Technion – Israel Institute of Technology.