GNNs Also Deserve Editing, and They Need It More Than Once

Shaochen Zhong, Duy Le, Zirui Liu, Zhimeng Jiang, Andrew Ye, Jiamu Zhang, Jiayi Yuan, Kaixiong Zhou, Zhaozhuo Xu, Jing Ma, Shuai Xu, Vipin Chaudhary, Xia Hu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:61727-61746, 2024.

Abstract

Suppose a self-driving car is crashing into pedestrians, or a chatbot is instructing its users to conduct criminal wrongdoing; the stakeholders of such products will undoubtedly want to patch these catastrophic errors as soon as possible. To address such concerns, Model Editing: the study of efficiently patching model behaviors without significantly altering their general performance, has seen considerable activity, with hundreds of editing techniques developed in various domains such as CV and NLP. However, the graph learning community has objectively fallen behind with only a few Graph Neural Network-compatible — and just one GNN-specific — model editing methods available, where all of which are limited in their practical scope. We argue that the impracticality of these methods lies in their lack of Sequential Editing Robustness: the ability to edit multiple errors sequentially, and therefore fall short in effectiveness, as this approach mirrors how errors are discovered and addressed in the real world. In this paper, we delve into the specific reasons behind the difficulty of editing GNNs in succession and observe the root cause to be model overfitting. We subsequently propose a simple yet effective solution — SEED-GNN — by leveraging overfit-prevention techniques in a GNN-specific context to derive the first and only GNN model editing method that scales practically. Additionally, we formally frame the task paradigm of GNN editing and hope to inspire future research in this crucial but currently overlooked field. Please refer to our GitHub repository for code and checkpoints.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-zhong24d, title = {{GNN}s Also Deserve Editing, and They Need It More Than Once}, author = {Zhong, Shaochen and Le, Duy and Liu, Zirui and Jiang, Zhimeng and Ye, Andrew and Zhang, Jiamu and Yuan, Jiayi and Zhou, Kaixiong and Xu, Zhaozhuo and Ma, Jing and Xu, Shuai and Chaudhary, Vipin and Hu, Xia}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {61727--61746}, 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/zhong24d/zhong24d.pdf}, url = {https://proceedings.mlr.press/v235/zhong24d.html}, abstract = {Suppose a self-driving car is crashing into pedestrians, or a chatbot is instructing its users to conduct criminal wrongdoing; the stakeholders of such products will undoubtedly want to patch these catastrophic errors as soon as possible. To address such concerns, Model Editing: the study of efficiently patching model behaviors without significantly altering their general performance, has seen considerable activity, with hundreds of editing techniques developed in various domains such as CV and NLP. However, the graph learning community has objectively fallen behind with only a few Graph Neural Network-compatible — and just one GNN-specific — model editing methods available, where all of which are limited in their practical scope. We argue that the impracticality of these methods lies in their lack of Sequential Editing Robustness: the ability to edit multiple errors sequentially, and therefore fall short in effectiveness, as this approach mirrors how errors are discovered and addressed in the real world. In this paper, we delve into the specific reasons behind the difficulty of editing GNNs in succession and observe the root cause to be model overfitting. We subsequently propose a simple yet effective solution — SEED-GNN — by leveraging overfit-prevention techniques in a GNN-specific context to derive the first and only GNN model editing method that scales practically. Additionally, we formally frame the task paradigm of GNN editing and hope to inspire future research in this crucial but currently overlooked field. Please refer to our GitHub repository for code and checkpoints.} }
Endnote
%0 Conference Paper %T GNNs Also Deserve Editing, and They Need It More Than Once %A Shaochen Zhong %A Duy Le %A Zirui Liu %A Zhimeng Jiang %A Andrew Ye %A Jiamu Zhang %A Jiayi Yuan %A Kaixiong Zhou %A Zhaozhuo Xu %A Jing Ma %A Shuai Xu %A Vipin Chaudhary %A Xia Hu %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-zhong24d %I PMLR %P 61727--61746 %U https://proceedings.mlr.press/v235/zhong24d.html %V 235 %X Suppose a self-driving car is crashing into pedestrians, or a chatbot is instructing its users to conduct criminal wrongdoing; the stakeholders of such products will undoubtedly want to patch these catastrophic errors as soon as possible. To address such concerns, Model Editing: the study of efficiently patching model behaviors without significantly altering their general performance, has seen considerable activity, with hundreds of editing techniques developed in various domains such as CV and NLP. However, the graph learning community has objectively fallen behind with only a few Graph Neural Network-compatible — and just one GNN-specific — model editing methods available, where all of which are limited in their practical scope. We argue that the impracticality of these methods lies in their lack of Sequential Editing Robustness: the ability to edit multiple errors sequentially, and therefore fall short in effectiveness, as this approach mirrors how errors are discovered and addressed in the real world. In this paper, we delve into the specific reasons behind the difficulty of editing GNNs in succession and observe the root cause to be model overfitting. We subsequently propose a simple yet effective solution — SEED-GNN — by leveraging overfit-prevention techniques in a GNN-specific context to derive the first and only GNN model editing method that scales practically. Additionally, we formally frame the task paradigm of GNN editing and hope to inspire future research in this crucial but currently overlooked field. Please refer to our GitHub repository for code and checkpoints.
APA
Zhong, S., Le, D., Liu, Z., Jiang, Z., Ye, A., Zhang, J., Yuan, J., Zhou, K., Xu, Z., Ma, J., Xu, S., Chaudhary, V. & Hu, X.. (2024). GNNs Also Deserve Editing, and They Need It More Than Once. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:61727-61746 Available from https://proceedings.mlr.press/v235/zhong24d.html.

Related Material