Rethinking Graph Super-Resolution: Dual Frameworks for Topological Fidelity

Pragya Singh, Islem Rekik

Keywords: Deep Graph Learning, Graph Super-resolution, Network Neuroscience

Abstract: Graph super-resolution is an underexplored yet highly relevant research direction that circumvents the need for costly and time-consuming data collection, preparation, and storage. This makes it especially desirable for resource-constrained fields such as the medical domain. Existing work on graph super-resolution leverages graph neural networks (GNNs) and achieves impressive results. However, we note two major limitations in the current model design: (1) It violates the underlying graph structure when increasing the number of nodes, and (2) it relies heavily on node representation learning, which has limited capacity to accurately model edges. To address these limitations, we propose two novel frameworks: (1) Bi-SR, which performs structure-aware node super-resolution, and (2) DEFEND, which focuses on edge representation learning for enhanced edge modeling. We supplement our work with rigorous theoretical analysis and conduct extensive experiments on simulated and real-world datasets covering diverse graph topologies and low-to-high resolution relationships. The results demonstrate substantial improvements across all experiments, highlighting the potential of both frameworks for graph super-resolution tasks.

Primary Area: learning on graphs and other geometries & topologies

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 3522