Research map / Graph learning / Graph self-supervised learning
Deep Relational Topic Modeling via Graph Poisson Gamma Belief Network
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 |
|---|---|---|---|
| add hierarchical (deep) topic layers to GPFA, giving GPGBN (ablation) | GPFA · Coil node clustering | ACC: 73.2 -> 73.6 | Helped |
| add hierarchical layers to GPFA (ablation) | GPFA · R8 node clustering | ACC: 72.4 -> 73.3 | Helped |
| use 3 stochastic layers instead of 1 in WGCAE (ablation) | WGCAE · Coil node clustering | ACC: 80.5 -> 83.3 | Helped |
| use 3 stochastic layers instead of 1 in WGAE (ablation) | WGAE · Coil node clustering | ACC: 77.5 -> 81.6 | Helped |
| use 3 stochastic layers instead of 1 in WGCAE (ablation) | WGCAE · R8 node clustering | ACC: 77.5 -> 78.2 | Helped |
| use 3 layers instead of 1 in WGCAE (ablation) | WGCAE · Cora link prediction | AUC: 93.4 -> 95.0 | Helped |
| use 3 layers instead of 1 in WGAE (ablation) | WGAE · Cora link prediction | AUC: 92.6 -> 93.8 | Helped |
| replace FNN encoder with GCN encoder (WGCAE vs WGAE) (ablation) | WGAE · Coil node clustering | ACC: 81.6 -> 83.3 | Helped |
| add node-feature generation to graph autoencoder (WGCAE vs VGAE) | VGAE · Coil node clustering | ACC: 66.3 -> 83.3 | Helped |
| increase beta (edge weight) toward 100 instead of 0.1 (ablation) | WGCAE · Coil link prediction vs node clustering | AUC/AP vs NMI/ACC | Mixed |
| WGCAE-layer3 vs SIG-VAE large K,J | WGCAE · Cora link prediction | AUC: 96.0 -> 95.0 | Hurt |
Closest papers by meaning
- Convolutional Poisson Gamma Belief Network (ICML 2019 · similarity 0.91)
- Sawtooth Factorial Topic Embeddings Guided Gamma Belief Network (ICML 2021 · similarity 0.89)
- Bidirectional Convolutional Poisson Gamma Dynamical Systems (NeurIPS 2020 · similarity 0.88)
- Topic Modeling Revisited: A Document Graph-based Neural Network Perspective (NeurIPS 2021 · similarity 0.87)
- Relational Deep Learning: A Deep Latent Variable Model for Link Prediction (AAAI 2017 · similarity 0.87)
- Generalization Principles for Inference over Text-Attributed Graphs with Large Language Models (ICML 2025 · similarity 0.84)
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.