Graph Mixing Additive Networks

2025-09-28

AI Summary

Addressing the challenge of jointly achieving flexibility, expressiveness, and interpretability in sparse time-series modeling, this paper proposes the Graph Neural Additive Model (GMAN) framework. GMAN extends additive modeling to graph-structured representations by explicitly encoding temporal trajectories via directed graphs and incorporating feature grouping and graph encoding priors to enable fine-grained interpretability at the feature, node, and graph levels—while supporting controllable trade-offs between expressiveness and interpretability. Technically, it integrates sparse time-series embedding, hierarchical feature grouping, and interpretability-aware regularization. Empirically, GMAN significantly outperforms strong non-interpretable baselines on real-world tasks—including mortality prediction from sparse blood test sequences and fake news detection—while generating high-quality, domain-aligned, and actionable explanations.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

We introduce GMAN, a flexible, interpretable, and expressive framework that extends Graph Neural Additive Networks (GNANs) to learn from sets of sparse time-series data. GMAN represents each time-dependent trajectory as a directed graph and applies an enriched, more expressive GNAN to each graph. It allows users to control the interpretability-expressivity trade-off by grouping features and graphs to encode priors, and it provides feature, node, and graph-level interpretability. On real-world datasets, including mortality prediction from blood tests and fake-news detection, GMAN outperforms strong non-interpretable black-box baselines while delivering actionable, domain-aligned explanations.

Problem

• Extends graph neural networks to learn from sparse time-series data
• Provides interpretability at feature, node, and graph levels simultaneously
• Outperforms black-box models while delivering domain-aligned explanations

Innovation

• Extends Graph Neural Additive Networks framework
• Represents time-series as directed graph structures
• Balances interpretability with feature grouping control