Research map / Graph learning / GNN architectures
Deep Collective Inference
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 LSTM collective inference (DCI) instead of PLEM+N2V | DCI · seven network datasets | BAE: up to 12% error reduction | Helped |
| use DCI instead of all competing methods | DCI · seven network datasets | BAE: 25% average error reduction | Helped |
| add class label predictions instead of stacking class probabilities as features (ablation) | DCI-WSB · smaller datasets | error reduction: 1.6% gain | Helped |
| initialize first-iteration predictions from DRI instead of random prior (ablation) | DCI, DCI-S, DCI-B · smaller datasets | error reduction: 0.82% gain (DCI-WSB 0.06%, DCI-S 1.78%, DCI-B 1.04%) | Helped |
| order neighbors by personalized page rank instead of random order (ablation) | DCI-WSB · smaller datasets | BAE | No clear effect |
| add collective inference over non-collective DRI (ablation) | DRI-WSB · smaller datasets | BAE | Helped |
| add label skew corrections (data augmentation and balanced cross entropy) (ablation) | DCI · imbalanced datasets | BAE | Helped |
| use DCI instead of PLEM+N2V on Amazon DVD 7500 at low training proportions | DCI · Amazon DVD 7500 | BAE | Mixed |
Closest papers by meaning
- A Collective Learning Framework to Boost GNN Expressiveness for Node Classification (ICML 2021 · similarity 0.91)
- Column Networks for Collective Classification (AAAI 2017 · similarity 0.88)
- Unitary Convolutions for Learning on Graphs and Groups (NeurIPS 2024 · similarity 0.86)
- Fully-inductive Node Classification on Arbitrary Graphs (ICLR 2025 · similarity 0.86)
- An Efficient Subgraph-Inferring Framework for Large-Scale Heterogeneous Graphs (AAAI 2024 · similarity 0.85)
- Node Identifiers: Compact, Discrete Representations for Efficient Graph Learning (ICLR 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.