Research map / Graph learning / GNN architectures
Probabilistic Learning on Graphs via Contextual Architectures
Davide Bacciu, Federico Errica, Alessio Micheli
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 |
|---|---|---|---|
| add bigram features to unigram vertex representation (ablation) | CGMM · IMDB-M | accuracy: 47.36 -> 49.73 | Improved |
| add bigram features to unigram vertex representation (ablation) | CGMM · COLLAB | accuracy: 73.3 -> 75.50 | Improved |
| add bigram features to unigram vertex representation (ablation) | CGMM · IMDB-B | accuracy: 71.07 -> 72.1 | Improved |
| add bigram features to unigram vertex representation (ablation) | CGMM · NCI1 | accuracy: 77.02 -> 77.80 | Improved |
| add bigram features to unigram vertex representation (ablation) | CGMM · D&D | accuracy: 77.35 -> 77.15 | No clear effect |
| condition each layer on all previous layers (switching parent) (ablation) | CGMM · graph classification benchmarks | accuracy: 77.15 -> 77.20 | No clear effect |
| condition each layer on all previous layers (switching parent) (ablation) | CGMM · COLLAB | accuracy: 75.50 -> 76.06 | No clear effect |
| condition each layer on all previous layers (switching parent) (ablation) | CGMM · PPI | Micro F1: 60.2 -> 58.4 | Hurt |
| increase number of layers (1 to 20) (ablation) | CGMM · NCI1 | accuracy | Improved |
| increase number of layers (1 to 20) (ablation) | CGMM · PROTEINS and D&D | accuracy | Hurt |
| deeper network with smaller hidden size instead of wider shallow network (ablation) | CGMM · NCI1 validation | accuracy: size 60 -> size 45 | Improved |
| train a pool of candidate Bayesian Networks per layer and select best (ablation) | CGMM · NCI1 | accuracy | No clear effect |
Closest papers by meaning
- Contextual Graph Markov Model: A Deep and Generative Approach to Graph Processing (ICML 2018 · similarity 0.92)
- Bayesian Graph Convolutional Neural Networks for Semi-Supervised Classification (AAAI 2019 · similarity 0.89)
- Semi-Supervised Classification with Graph Convolutional Networks (ICLR 2017 · similarity 0.89)
- The Infinite Contextual Graph Markov Model (ICML 2022 · similarity 0.88)
- Learning Graph Representation via Graph Entropy Maximization (ICML 2024 · similarity 0.87)
- Towards characterizing the value of edge embeddings in Graph Neural Networks (ICML 2025 · 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.
