Projects per year
Abstract
Graph neural networks (GNNs) have been analyzed from multiple perspectives, including the WL-hierarchy, which exposes limits on their expressivity to distinguish graphs. However, characterizing the class of functions that they learn has remained unresolved. We address this fundamental question for message passing GNNs under ReLU activations, i.e., the de-facto choice for most GNNs.We first show that such GNNs learn tropical rational signomial maps or continuous piecewise linear functions, establishing an equivalence with feedforward networks (FNNs). We then elucidate the role of the choice of aggregation and update functions, and derive the first general upper and lower bounds on the geometric complexity (i.e., the number of linear regions), establishing new results for popular architectures such as GraphSAGE and GIN. We also introduce and theoretically analyze several new architectures to illuminate the relative merits of the feedforward and the message passing layers, and the tradeoffs involving depth and number of trainable parameters. Finally, we also characterize the decision boundary for node and graph classification tasks.
Original language | English |
---|---|
Title of host publication | Advances in Neural Information Processing Systems 37 (NeurIPS 2024) |
Editors | A. Globerson, L. Mackey, D. Belgrave, A. Fan, U. Paquet, J. Tomczak, C. Zhang |
Publisher | Curran Associates Inc. |
ISBN (Print) | 9798331314385 |
Publication status | Published - 2025 |
MoE publication type | A4 Conference publication |
Event | Conference on Neural Information Processing Systems - Vancouver, Canada, Vancouver , Canada Duration: 10 Dec 2024 → 15 Dec 2024 Conference number: 38 https://neurips.cc/Conferences/2024 |
Publication series
Name | Advances in Neural Information Processing Systems |
---|---|
Publisher | Curran Associates, Inc. |
Volume | 37 |
ISSN (Print) | 1049-5258 |
Conference
Conference | Conference on Neural Information Processing Systems |
---|---|
Abbreviated title | NeurIPS |
Country/Territory | Canada |
City | Vancouver |
Period | 10/12/2024 → 15/12/2024 |
Internet address |
Fingerprint
Dive into the research topics of 'What do Graph Neural Networks learn? Insights from Tropical Geometry'. Together they form a unique fingerprint.Projects
- 1 Active
-
HEALED/Garg: Human-steered next-generation machine learning for reviving drug design
Garg, V. (Principal investigator)
01/09/2021 → 31/08/2025
Project: RCF Academy Project