Research map / Graph learning / Graph self-supervised learning
Deconvolutional Networks on Graph Data
Jia Li, Jiajin Li, Yang Liu, Jianwei Yu, Yueting Li, Hong Cheng
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 |
|---|---|---|---|
| Use GDN instead of GRAPE | Graph feature imputation · Ciao, Douban, Cora, Citeseer, Amaphoto, Amacomp | RMSE: 1.042 -> 1.011 (Ciao); 0.742 -> 0.734 (Douban); 0.430 -> 0.415 (Cora); 0.419 -> 0.399 (Citeseer); 0.399 -> 0.391 (Amapho | Improved |
| Add GDN feature reconstruction | VGAE · MUTAG, PTC-MR, ZINC | AUC: 0.716 -> 0.823 (MUTAG); 0.566 -> 0.760 (PTC-MR); 0.556 -> 0.858 (ZINC) | Improved |
| Add GDN feature reconstruction | Graphite · MUTAG, PTC-MR, ZINC | AP: 0.447 -> 0.608 (MUTAG); 0.437 -> 0.613 (PTC-MR); 0.288 -> 0.567 (ZINC) | Improved |
| Add wavelet-domain denoising after the inverse filter (ablation) | GDN graph feature imputation decoder · Ciao, Douban, Cora, Citeseer, Amaphoto, Amacomp | RMSE: Inverse decoder: 1.115, 0.812, 0.447, 0.431, 0.409, 0.407 -> OURS: 1.011, 0.734, 0.415, 0.399, 0.391, 0.393 | Improved |
| Use the third-order rather than first-order Maclaurin approximation (ablation) | GDN with VGAE · MUTAG, PTC-MR, ZINC | AUC: 0.738 -> 0.823 (MUTAG); 0.493 -> 0.760 (PTC-MR); 0.474 -> 0.858 (ZINC) | Improved |
| Use the third-order rather than first-order Maclaurin approximation (ablation) | GDN with Graphite · MUTAG, PTC-MR, ZINC | AUC: 0.741 -> 0.818 (MUTAG); 0.529 -> 0.773 (PTC-MR); 0.478 -> 0.838 (ZINC) | Improved |
| Use the third-order inverse approximation rather than GALA's first-order approximation (ablation) | Graph feature imputation decoder · Ciao, Douban, Cora, Citeseer, Amaphoto, Amacomp | RMSE: Inverse decoder: 1.115, 0.812, 0.447, 0.431, 0.409, 0.407; GALA: 1.147, 0.833, 0.457, 0.435, 0.415, 0.411 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Graphite — Graphite: Iterative Generative Modeling of Graphs (ICML 2019) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Simplifying Graph Convolutional Networks (ICML 2019 · similarity 0.90)
- Wiener Graph Deconvolutional Network Improves Graph Self-Supervised Learning (AAAI 2023 · similarity 0.89)
- Graph Wavelet Neural Network (ICLR 2019 · similarity 0.88)
- WaveNet: Tackling Non-stationary Graph Signals via Graph Spectral Wavelets (AAAI 2024 · similarity 0.87)
- A General Graph Spectral Wavelet Convolution via Chebyshev Order Decomposition (ICML 2025 · similarity 0.87)
- HoloNets: Spectral Convolutions do extend to Directed Graphs (ICLR 2024 · similarity 0.87)
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.
