Research map / Graph learning / Graph self-supervised learning
Graph-Based Semi-Supervised Learning with Non-ignorable Non-response
Fan Zhou, Tengfei Li, Haibo Zhou, Hongtu Zhu, Ye Jieping
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| model non-ignorable missingness (GNM) instead of standard GCN (SM) | GNM · simulated data p=4, sigma=0.5 | RMSE: 1.1925 -> 0.6983 | Improved |
| model non-ignorable missingness (GNM) instead of standard GCN (SM) | GNM · simulated data p=16, sigma=1 | RMSE: 1.4212 -> 1.1316 | Improved |
| model non-ignorable missingness (GNM) instead of standard GCN (SM) | GNM · simulated data p=4, sigma=1 | MAPE: 0.3104 -> 0.2263 | Improved |
| use GNM instead of SM | GNM · Cora binary simple setup lambda=1 | accuracy: 0.8683 -> 0.8947 | Improved |
| use GNM instead of SM | GNM · Cora binary simple setup lambda=2 | accuracy: 0.8052 -> 0.8648 | Improved |
| use GNM instead of inverse-weighting method of Rosset | GNM · Cora binary simple setup lambda=1.5 | accuracy: 0.8311 -> 0.8908 | Improved |
| increase label imbalance lambda from 1 to 2 (SM) | SM · Cora binary simple setup | accuracy: 0.8683 -> 0.8052 | Hurt |
| increase label imbalance lambda from 1 to 2 (GNM) | GNM · Cora binary simple setup | accuracy: 0.8947 -> 0.8648 | Hurt |
| model non-ignorable missingness correctly instead of treating it as ignorable (GIM) (ablation) | GNM · Cora binary complicated setup lambda=1 | accuracy: 0.8713 -> 0.8961 | Improved |
| model non-ignorable missingness correctly instead of treating it as ignorable (GIM) (ablation) | GNM · Cora binary complicated setup lambda=2 | accuracy: 0.8291 -> 0.8669 | Improved |
| treat missingness as ignorable (GIM) instead of ignoring missingness altogether (SM) (ablation) | GNM · Cora binary complicated setup lambda=2 | accuracy: 0.8141 -> 0.8291 | Improved |
| use GNM instead of Rosset in terms of variance | GNM · Cora binary simple setup lambda=1 | accuracy SD: 5.19e-2 -> 6.47e-3 | Improved |
Closest papers by meaning
- Unlabeled Data Help in Graph-Based Semi-Supervised Learning: A Bayesian Nonparametrics Perspective (JMLR 2022 · similarity 0.87)
- On Non-Random Missing Labels in Semi-Supervised Learning (ICLR 2022 · similarity 0.87)
- A Flexible Generative Framework for Graph-based Semi-supervised Learning (NeurIPS 2019 · similarity 0.86)
- Label Efficient Semi-Supervised Learning via Graph Filtering (CVPR 2019 · similarity 0.86)
- Towards Semi-supervised Learning with Non-random Missing Labels (ICCV 2023 · similarity 0.86)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
