Research map / Graph learning / Graph self-supervised learning
Bag Graph: Multiple Instance Learning Using Bayesian Graph Neural Networks
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 |
|---|---|---|---|
| Replace the k-NN observed graph with Bayesian inferred graph learning | Res+pool-GCN on 20 Newsgroups · 20 Newsgroups (20 datasets) | average rank: 4.55 -> 3.35 | Helped |
| Use the k-NN observed graph rather than a graph-agnostic model | Res+pool on 20 Newsgroups · 20 Newsgroups (20 datasets) | average rank: 4.05 -> 4.55 | Hurt |
| Add spatial graph convolution | Deep Sets · 2016 US election counties | accuracy: 73.22±3.22 -> 74.05±4.56 | Helped |
| Replace the observed spatial graph with Bayesian inferred graph learning | DS-GCN · 2016 US election counties | accuracy: 74.05±4.56 -> 74.29±3.15 | Helped |
| Add graph convolution using the observed neighborhood graph | Deep Sets · NYC rental prices | RMSE: 86.37±20.41 -> 78.57±16.06 | Helped |
| Replace observed graph convolution with Bayesian inferred graph learning | DS-GCN · NYC rental prices | RMSE: 78.57±16.06 -> 67.51±16.39 | Helped |
| Replace observed graph convolution with Bayesian inferred graph learning | ST-GCN · NYC rental prices | RMSE: 71.86±14.65 -> 69.44±16.23 | Helped |
| Perform graph inference only on training nodes and add disconnected test nodes at evaluation (ablation) | B-DS-GCN · NYC rental prices (DS) | RMSE: 67.51±16.39 -> 75.26±16.99 | Hurt |
| Train on training nodes only, but infer graph using training and test nodes at evaluation (ablation) | B-DS-GCN · NYC rental prices (DS) | RMSE: 67.51±16.39 -> 68.15±16.77 | Hurt |
| Perform graph inference and model training transductively on training and test nodes (ablation) | B-ST-GCN · NYC rental prices (ST) | RMSE: 89.95±23.23 -> 69.44±16.23 | Helped |
| Train on training nodes only, but infer graph using training and test nodes at evaluation (ablation) | B-ST-GCN · NYC rental prices (ST) | RMSE: 69.44±16.23 -> 71.72±16.50 | Hurt |
Closest papers by meaning
- Robust Self-Supervised Multi-Instance Learning with Structure Awareness (AAAI 2023 · similarity 0.89)
- Rethinking Multi-Instance Learning Through Graph-Driven Fusion: A Dual-Path Approach to Adaptive Representation (AAAI 2026 · similarity 0.88)
- Variational Bayesian Multiple Instance Learning With Gaussian Processes (CVPR 2017 · similarity 0.86)
- Multi-Instance Causal Representation Learning for Instance Label Prediction and Out-of-Distribution Generalization (NeurIPS 2022 · similarity 0.86)
- Attention-based Deep Multiple Instance Learning (ICML 2018 · similarity 0.86)
- Sm: enhanced localization in Multiple Instance Learning for medical imaging classification (NeurIPS 2024 · 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.