Graph Additive Models disentangling feature, topology, and interactions
Abstract
Graph Neural Additive Networks (GNANs) have emerged as effective, interpretable-by-design architectures for graph learning. They achieve transparency by combining per-feature shape functions with simple distance-based reweighting. However, GNANs have three limitations. First, node-wise relevance scores aggregate information from all graph nodes, allowing irrelevant nodes to inherit spurious relevance from relevant ones. Second, topology is represented only through shortest-path distances, so structurally distinct graphs may share identical distance profiles. Third, feature and distance information are combined multiplicatively, entangling feature, distance, and interaction effects in a single rescalable term. We introduce AMIGA, which represents topology through a vocabulary of mined motifs and decomposes each node contribution into three additive channels: feature, structure, and feature-structure interaction. The channels are centered around a graph-balanced training baseline, making the decomposition well-defined and unique under the chosen reference distribution. Across ten graph classification and regression benchmarks, AMIGA outperforms GNAN on most datasets and narrows the gap with standard black-box GNNs. Because each channel is centered and assigned to a single node, AMIGA produces node-local explanations that reconstruct the model output exactly and support interpretation at multiple levels of granularity.
Related
- Diameter and radius of an additive-multiplicative graph
- Additive Quasi-isometry via rooted graph partitions and layering partition
- Graphic additive bases and graph flows
- From topology learning to graph generation: A unifying perspective
- Extremal List Gaps and Inapproximability in Additive Graph Labeling
BibTeX
@misc{alvarezrodriguez2026graph,
title = {{Graph Additive Models disentangling feature, topology, and interactions}},
author = {Clara Alvarez-Rodriguez and Steve Azzolin and Vincenzo Marco De Luca and Filippo Maria Bianchi and Andrea Passerini and Antonio Longa},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.20662},
eprint = {2610.20662},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.20662},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Graph Additive Models disentangling feature, topology, and interactions
AU - Clara Alvarez-Rodriguez
AU - Steve Azzolin
AU - Vincenzo Marco De Luca
AU - Filippo Maria Bianchi
AU - Andrea Passerini
AU - Antonio Longa
PY - 2026
DA - 2026/10/14
PB - arXiv
DO - 10.48550/arxiv.2610.20662
UR - https://arxiv.org/abs/2610.20662
AB - Graph Neural Additive Networks (GNANs) have emerged as effective, interpretable-by-design architectures for graph learning. They achieve transparency by combining per-feature shape functions with simple distance-based reweighting. However, GNANs have three limitations. First, node-wise relevance scores aggregate information from all graph nodes, allowing irrelevant nodes to inherit spurious relevance from relevant ones. Second, topology is represented only through shortest-path distances, so structurally distinct graphs may share identical distance profiles. Third, feature and distance information are combined multiplicatively, entangling feature, distance, and interaction effects in a single rescalable term. We introduce AMIGA, which represents topology through a vocabulary of mined motifs and decomposes each node contribution into three additive channels: feature, structure, and feature-structure interaction. The channels are centered around a graph-balanced training baseline, making the decomposition well-defined and unique under the chosen reference distribution. Across ten graph classification and regression benchmarks, AMIGA outperforms GNAN on most datasets and narrows the gap with standard black-box GNNs. Because each channel is centered and assigned to a single node, AMIGA produces node-local explanations that reconstruct the model output exactly and support interpretation at multiple levels of granularity.
ER -