Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2508.02044 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916879208022016 |
|---|---|
| author | Yin, Hang Liu, Zipeng Peng, Xiaoyong Xiang, Liyao |
| author_facet | Yin, Hang Liu, Zipeng Peng, Xiaoyong Xiang, Liyao |
| contents | Graph unlearning is tailored for GNNs to handle widespread and various graph structure unlearning requests, which remain largely unexplored. The GIF (graph influence function) achieves validity under partial edge unlearning, but faces challenges in dealing with more disturbing node unlearning. To avoid the overhead of retraining and realize the model utility of unlearning, we proposed a novel node unlearning method to reverse the process of aggregation in GNN by embedding reconstruction and to adopt Range-Null Space Decomposition for the nodes' interaction learning. Experimental results on multiple representative datasets demonstrate the SOTA performance of our proposed approach. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_02044 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Graph Unlearning via Embedding Reconstruction -- A Range-Null Space Decomposition Approach Yin, Hang Liu, Zipeng Peng, Xiaoyong Xiang, Liyao Machine Learning Artificial Intelligence Social and Information Networks 68T05 K.4.1 Graph unlearning is tailored for GNNs to handle widespread and various graph structure unlearning requests, which remain largely unexplored. The GIF (graph influence function) achieves validity under partial edge unlearning, but faces challenges in dealing with more disturbing node unlearning. To avoid the overhead of retraining and realize the model utility of unlearning, we proposed a novel node unlearning method to reverse the process of aggregation in GNN by embedding reconstruction and to adopt Range-Null Space Decomposition for the nodes' interaction learning. Experimental results on multiple representative datasets demonstrate the SOTA performance of our proposed approach. |
| title | Graph Unlearning via Embedding Reconstruction -- A Range-Null Space Decomposition Approach |
| topic | Machine Learning Artificial Intelligence Social and Information Networks 68T05 K.4.1 |
| url | https://arxiv.org/abs/2508.02044 |