Editing Partially Observable Networks via Graph Diffusion Models

Puja Trivedi, Ryan A. Rossi, David Arbour, Tong Yu, Franck Dernoncourt, Sungchul Kim, Nedim Lipka, Namyong Park, Nesreen K. Ahmed, Danai Koutra
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:48608-48629, 2024.

Abstract

Most real-world networks are noisy and incomplete samples from an unknown target distribution. Refining them by correcting corruptions or inferring unobserved regions typically improves downstream performance. Inspired by the impressive generative capabilities that have been used to correct corruptions in images, and the similarities between "in-painting" and filling in missing nodes and edges conditioned on the observed graph, we propose a novel graph generative framework, SGDM, which is based on subgraph diffusion. Our framework not only improves the scalability and fidelity of graph diffusion models, but also leverages the reverse process to perform novel, conditional generation tasks. In particular, through extensive empirical analysis and a set of novel metrics, we demonstrate that our proposed model effectively supports the following refinement tasks for partially observable networks: (T1) denoising extraneous subgraphs, (T2) expanding existing subgraphs and (T3) performing “style" transfer by regenerating a particular subgraph to match the characteristics of a different node or subgraph.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-trivedi24a, title = {Editing Partially Observable Networks via Graph Diffusion Models}, author = {Trivedi, Puja and Rossi, Ryan A. and Arbour, David and Yu, Tong and Dernoncourt, Franck and Kim, Sungchul and Lipka, Nedim and Park, Namyong and Ahmed, Nesreen K. and Koutra, Danai}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {48608--48629}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/trivedi24a/trivedi24a.pdf}, url = {https://proceedings.mlr.press/v235/trivedi24a.html}, abstract = {Most real-world networks are noisy and incomplete samples from an unknown target distribution. Refining them by correcting corruptions or inferring unobserved regions typically improves downstream performance. Inspired by the impressive generative capabilities that have been used to correct corruptions in images, and the similarities between "in-painting" and filling in missing nodes and edges conditioned on the observed graph, we propose a novel graph generative framework, SGDM, which is based on subgraph diffusion. Our framework not only improves the scalability and fidelity of graph diffusion models, but also leverages the reverse process to perform novel, conditional generation tasks. In particular, through extensive empirical analysis and a set of novel metrics, we demonstrate that our proposed model effectively supports the following refinement tasks for partially observable networks: (T1) denoising extraneous subgraphs, (T2) expanding existing subgraphs and (T3) performing “style" transfer by regenerating a particular subgraph to match the characteristics of a different node or subgraph.} }
Endnote
%0 Conference Paper %T Editing Partially Observable Networks via Graph Diffusion Models %A Puja Trivedi %A Ryan A. Rossi %A David Arbour %A Tong Yu %A Franck Dernoncourt %A Sungchul Kim %A Nedim Lipka %A Namyong Park %A Nesreen K. Ahmed %A Danai Koutra %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-trivedi24a %I PMLR %P 48608--48629 %U https://proceedings.mlr.press/v235/trivedi24a.html %V 235 %X Most real-world networks are noisy and incomplete samples from an unknown target distribution. Refining them by correcting corruptions or inferring unobserved regions typically improves downstream performance. Inspired by the impressive generative capabilities that have been used to correct corruptions in images, and the similarities between "in-painting" and filling in missing nodes and edges conditioned on the observed graph, we propose a novel graph generative framework, SGDM, which is based on subgraph diffusion. Our framework not only improves the scalability and fidelity of graph diffusion models, but also leverages the reverse process to perform novel, conditional generation tasks. In particular, through extensive empirical analysis and a set of novel metrics, we demonstrate that our proposed model effectively supports the following refinement tasks for partially observable networks: (T1) denoising extraneous subgraphs, (T2) expanding existing subgraphs and (T3) performing “style" transfer by regenerating a particular subgraph to match the characteristics of a different node or subgraph.
APA
Trivedi, P., Rossi, R.A., Arbour, D., Yu, T., Dernoncourt, F., Kim, S., Lipka, N., Park, N., Ahmed, N.K. & Koutra, D.. (2024). Editing Partially Observable Networks via Graph Diffusion Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:48608-48629 Available from https://proceedings.mlr.press/v235/trivedi24a.html.

Related Material