Research map / Architectures and efficiency / Neural architecture search
Relatron: Automating Relational Machine Learning over Relational Databases
Zhikai Chen, Han Xie, Jian Zhang, Jiliang Tang, Xiang song, Huzefa Rangwala
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 task-embedding meta-predictor (model-free features) for choosing RDL vs DFS (ablation) | Relatron · RDL vs DFS winner prediction | LOO accuracy: 66.7% -> 87.5% | Improved |
| increase performance-bank size from 8 to all tasks (ablation) | model-free meta-predictor · RDL vs DFS winner prediction | LOO accuracy: 68.9% -> 87.5% | Improved |
| select checkpoint by test instead of validation | RDL on driver-top3 · driver-top3 | Test AUROC: 82.41 -> 85.94 | Improved |
| select RelGNN by validation instead of test | RelGNN · driver-top3 | AUROC: 85.69 -> 82.61 | Hurt |
| select GraphSAGE by validation instead of test | GraphSAGE · user-ignore | AUROC: 86.40 -> 72.27 | Hurt |
| add Relatron meta-predictor to TPE search at budget 30 | TPE · driver-top3 | ROC-AUC: 77.92 -> 84.00 | Improved |
| add Relatron meta-predictor to TPE search at budget 30 | TPE · driver-position | MAE: 4.1724 -> 3.3339 | Improved |
| add post-hoc landscape-based checkpoint selection (full vs only predictor) at budget 30 (ablation) | Relatron · driver-top3 | ROC-AUC: 84.00 -> 84.33 | No clear effect |
| increase search budget from 3 to 30 trials | Random search · driver-top3 | ROC-AUC: 82.67 -> 77.28 | Hurt |
| increase search budget from 3 to 30 trials | Relatron (only predictor) · driver-top3 | ROC-AUC: 83.80 -> 84.00 | No clear effect |
| add Autotransfer embedding-based retrieval | TPE · driver-top3 | ROC-AUC: 81.45 -> 78.71 | Hurt |
| add Autotransfer embedding-based retrieval | TPE · user-churn | ROC-AUC: 68.60 -> 70.28 | Improved |
Closest papers by meaning
- Position: Relational Deep Learning - Graph Representation Learning on Relational Databases (ICML 2024 · similarity 0.89)
- RelGNN: Composite Message Passing for Relational Deep Learning (ICML 2025 · similarity 0.87)
- Griffin: Towards a Graph-Centric Relational Database Foundation Model (ICML 2025 · similarity 0.87)
- RDB2G-Bench: A Comprehensive Benchmark for Automatic Graph Modeling of Relational Databases (NeurIPS 2025 · similarity 0.86)
- A Pre-training Framework for Relational Data with Information-theoretic Principles (NeurIPS 2025 · similarity 0.85)
- 4DBInfer: A 4D Benchmarking Toolbox for Graph-Centric Predictive Modeling on RDBs (NeurIPS 2024 · 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.
