Boosting reinforcement learning for critical node detection with multi-topology training strategies

Identifying pivotal components in complex networks—such as key intersections in transportation systems, influential actors in social networks, or keystone species in ecosystems—is essential for improving the resilience and mitigating catastrophic failures of these systems. However, existing methods for critical node detection are typically tailored to specific network types and objectives, limiting their generalizability. To overcome this limitation, we propose a machine learning framework that combines reinforcement learning with an attention-based graph neural network. By leveraging multi-topology training, our model learns to adapt and generalize across diverse network structures and objective functions. Experiments on real-world and synthetic networks uncover distinct topological features that are important for identifying critical nodes. Notably, our method applied to personalized gene networks, prioritizes genes that are significantly enriched for cataloged cancer drivers. This work provides a robust critical node detection model with a comprehensive multi-topology training approach, suitable for a broader range of real-world applications. Critical node detection helps identify network elements whose removal most disrupts connectivity. Here, the authors develop a reinforcement-learning approach that trains on diverse graph topologies to improve generalization to both synthetic and real networks.

Authors

Institutions

Publication Details

Journal
Nature Communications
Published
2026-11-14
DOI
https://doi.org/10.1038/s41467-026-78707-2
Primary Topic
Complex Network Analysis Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Boosting reinforcement learning for critical node detection with multi-topology training strategies

Jianxi Gao, Pin‐Yu Chen, Shlomo Havlin, Kushal Raj Bhandari
Nature Communications
Complex Network Analysis Techniques
article

Boosting reinforcement learning for critical node detection with multi-topology training strategies

Jianxi Gao, Pin‐Yu Chen, Shlomo Havlin, Kushal Raj Bhandari
article en

Abstract

Identifying pivotal components in complex networks—such as key intersections in transportation systems, influential actors in social networks, or keystone species in ecosystems—is essential for improving the resilience and mitigating catastrophic failures of these systems. However, existing methods for critical node detection are typically tailored to specific network types and objectives, limiting their generalizability. To overcome this limitation, we propose a machine learning framework that combines reinforcement learning with an attention-based graph neural network. By leveraging multi-topology training, our model learns to adapt and generalize across diverse network structures and objective functions. Experiments on real-world and synthetic networks uncover distinct topological features that are important for identifying critical nodes. Notably, our method applied to personalized gene networks, prioritizes genes that are significantly enriched for cataloged cancer drivers. This work provides a robust critical node detection model with a comprehensive multi-topology training approach, suitable for a broader range of real-world applications. Critical node detection helps identify network elements whose removal most disrupts connectivity. Here, the authors develop a reinforcement-learning approach that trains on diverse graph topologies to improve generalization to both synthetic and real networks.

Nature Communications
IBM (United States) (US), Bar-Ilan University (IL), Rensselaer Polytechnic Institute (US), IBM Research - Thomas J. Watson Research Center (US)
Complex Network Analysis Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon