HARMONIA: Interpretable Graph Learning through Mixtures of Neural Bases
Abstract
Existing interpretable graph additive models still face limitations in either computational scalability or modeling flexibility. In terms of structural modeling, previous approaches either face quadratic scaling costs or sacrifice explicit source-to-target contribution decomposition. In terms of feature components, they rely either on per-feature neural networks or on single shared bases with limited feature specialization. We address both problems by introducing HARMONIA: Interpretable Graph Learning through Mixtures of Neural Bases, an interpretable-by-design framework. For feature modeling, HARMONIA introduces a Mixture of Neural Bases (MoNB), which routes features to specialized basis experts, enabling parameter sharing without sacrificing feature-specific specialization. For structural modeling, HARMONIA uses Relative Random Walk Probabilities (RRWP) to capture multi-hop and multi-path relationships, and proposes Sparse RRWP Aggregation (SRA) to compute these interactions through sparse graph propagation without quadratic pairwise complexity. HARMONIA retains a simple additive form in which predictions decompose into feature responses modulated by structural influence. Empirically, HARMONIA achieves stronger explanation recovery than existing interpretable graph baselines while maintaining competitive predictive performance and scaling to graphs with millions of nodes. These results show that interpretable graph learning can remain both faithful and scalable without sacrificing predictive effectiveness.
Related
- Dimension reduction of compositional data through mixtures of latent source compositions
- Formulating Cross-World Mediation Estimands Through Single-World Mixtures
- Bits per Spike as a Betting Game: An Interpretable Unit for Held-Out Log-Likelihood in Neural Data Analysis
- Interpretable intrinsic dimension estimation through componentwise calibration of distance and angle
- Mastering Model Fit: Applications of the InterModel Vigorish for Interpretable Machine Learning
BibTeX
@misc{bui2026harmonia,
title = {{HARMONIA: Interpretable Graph Learning through Mixtures of Neural Bases}},
author = {Quan D. Bui and Nguyen Do and An Nguyen Dang and Huyen Nguyen and Nhu Duc Minh Nguyen and My T. Thai},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.33972},
eprint = {2609.33972},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.33972},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - HARMONIA: Interpretable Graph Learning through Mixtures of Neural Bases
AU - Quan D. Bui
AU - Nguyen Do
AU - An Nguyen Dang
AU - Huyen Nguyen
AU - Nhu Duc Minh Nguyen
AU - My T. Thai
PY - 2026
DA - 2026/09/27
PB - arXiv
DO - 10.48550/arxiv.2609.33972
UR - https://arxiv.org/abs/2609.33972
AB - Existing interpretable graph additive models still face limitations in either computational scalability or modeling flexibility. In terms of structural modeling, previous approaches either face quadratic scaling costs or sacrifice explicit source-to-target contribution decomposition. In terms of feature components, they rely either on per-feature neural networks or on single shared bases with limited feature specialization. We address both problems by introducing HARMONIA: Interpretable Graph Learning through Mixtures of Neural Bases, an interpretable-by-design framework. For feature modeling, HARMONIA introduces a Mixture of Neural Bases (MoNB), which routes features to specialized basis experts, enabling parameter sharing without sacrificing feature-specific specialization. For structural modeling, HARMONIA uses Relative Random Walk Probabilities (RRWP) to capture multi-hop and multi-path relationships, and proposes Sparse RRWP Aggregation (SRA) to compute these interactions through sparse graph propagation without quadratic pairwise complexity. HARMONIA retains a simple additive form in which predictions decompose into feature responses modulated by structural influence. Empirically, HARMONIA achieves stronger explanation recovery than existing interpretable graph baselines while maintaining competitive predictive performance and scaling to graphs with millions of nodes. These results show that interpretable graph learning can remain both faithful and scalable without sacrificing predictive effectiveness.
ER -