Relatron: Automating Relational Machine Learning over Relational Databases

2026-02-25

AI Summary

This work addresses the challenges in predictive modeling over relational databases, where cross-table dependencies and feature interactions are difficult to model effectively, and there is a lack of systematic understanding of performance differences between relational deep learning (RDL) and traditional methods like Deep Feature Synthesis (DFS), as well as task-adaptive selection mechanisms. To bridge this gap, the authors propose a unified design space encompassing both RDL and DFS, construct a performance repository via architecture-centric search, and introduce two task-aware signals—task homogeneity and affinity embedding—to explain performance disparities. Building on these insights, they develop Relatron, a lightweight meta-selector enhanced with loss landscape flatness for improved robustness. Experiments demonstrate that Relatron achieves up to an 18.5% performance gain over strong baselines in joint hyperparameter and architecture optimization, at a computational cost ten times lower than Fisher information–based approaches, effectively mitigating the “more tuning, worse performance” phenomenon.

Technology Category

• Machine Learning: Statistical Relational/Logic Learning
• Search and Optimization: Metareasoning and Metaheuristics
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Predictive modeling over relational databases (RDBs) powers applications, yet remains challenging due to capturing both cross-table dependencies and complex feature interactions. Relational Deep Learning (RDL) methods automate feature engineering via message passing, while classical approaches like Deep Feature Synthesis (DFS) rely on predefined non-parametric aggregators. Despite performance gains, the comparative advantages of RDL over DFS and the design principles for selecting effective architectures remain poorly understood. We present a comprehensive study that unifies RDL and DFS in a shared design space and conducts architecture-centric searches across diverse RDB tasks. Our analysis yields three key findings: (1) RDL does not consistently outperform DFS, with performance being highly task-dependent; (2) no single architecture dominates across tasks, underscoring the need for task-aware model selection; and (3) validation accuracy is an unreliable guide for architecture choice. This search yields a model performance bank that links architecture configurations to their performance; leveraging this bank, we analyze the drivers of the RDL-DFS performance gap and introduce two task signals -- RDB task homophily and an affinity embedding that captures size, path, feature, and temporal structure -- whose correlation with the gap enables principled routing. Guided by these signals, we propose Relatron, a task embedding-based meta-selector that chooses between RDL and DFS and prunes the within-family search. Lightweight loss-landscape metrics further guard against brittle checkpoints by preferring flatter optima. In experiments, Relatron resolves the "more tuning, worse performance" effect and, in joint hyperparameter-architecture optimization, achieves up to 18.5% improvement over strong baselines with 10x lower cost than Fisher information-based alternatives.

Problem

• Relational Machine Learning
• Relational Databases
• Feature Engineering
• Model Selection
• Architecture Design

Innovation

• Relational Deep Learning
• Deep Feature Synthesis
• Task Embedding
• Architecture Selection
• Meta-learning