analiza aiqbench
MAVN introduces a differentiable framework for Message Passing Neural Networks that adaptively determines virtual node connectivity based on layer-specific importance, overcoming rigid constraints in existing architectures. Theoretically proven to simulate any connectivity pattern, the method achieves significant performance gains across nine datasets by dynamically selecting and connecting virtual nodes to optimize graph representation. This approach offers a flexible mechanism for enhancing backbone models without requiring fixed connection structures or pre-determined topologies.
Learn When and Where to Connect: Adaptive Virtual Nodes for Dynamic Message Passing on Graphs
arXiv:2606.03068v1 Announce Type: cross Abstract: While Virtual Nodes (VNs) are often utilized in Message Passing Neural Networks (MPNNs) to facilitate effective message passing, existing VN-based methods have limitations, such as constraining all nodes to connect to the same number of VNs, fixing the connections before applying MPNNs, and connecting a node to a VN independently of the other nodes that connect to the same VN. We propose MAVN, an end-to-end differentiable MPNN framework that allows non-constrained connections between nodes and VNs and dynamically introduces VNs on demand in response to evolving node representations across layers. Specifically, MAVN learns to adaptively determine when (at which layer) and where (to which nodes) to introduce and connect VNs based on the relative importance of connections. From a pool of candidate VNs, MAVN selects the necessary VNs in each layer, where each selected VN is connected to a nonempty subset of nodes, guided by a dual-perspective scoring mechanism that jointly captures the nodes' preferences for VNs and the VNs' preferences for nodes. We theoretically prove that for any node-VN connectivity pattern, there exists a set of MAVN's parameters that can simulate the pattern. Experiments on nine real-world datasets demonstrate that MAVN consistently improves the performance of backbone MPNNs, achieving up to 46.5% improvement over the backbones and outperforms the baselines.
Abstract:While Virtual Nodes (VNs) are often utilized in Message Passing Neural Networks (MPNNs) to facilitate effective message passing, existing VN-based methods have limitations, such as constraining all nodes to connect to the same number of VNs, fixing the connections before applying MPNNs, and connecting a node to a VN independently of the other nodes that connect to the same VN. We propose MAVN, an end-to-end differentiable MPNN framework that allows non-constrained connections between nodes and VNs and dynamically introduces VNs on demand in response to evolving node representations across layers. Specifically, MAVN learns to adaptively determine when (at which layer) and where (to which nodes) to introduce and connect VNs based on the relative importance of connections. From a pool of candidate VNs, MAVN selects the necessary VNs in each layer, where each selected VN is connected to a nonempty subset of nodes, guided by a dual-perspective scoring mechanism that jointly captures the nodes' preferences for VNs and the VNs' preferences for nodes. We theoretically prove that for any node-VN connectivity pattern, there exists a set of MAVN's parameters that can simulate the pattern. Experiments on nine real-world datasets demonstrate that MAVN consistently improves the performance of backbone MPNNs, achieving up to 46.5% improvement over the backbones and outperforms the baselines.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2606.03068 [cs.LG]
(or arXiv:2606.03068v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.03068 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
https://doi.org/10.1145/3770855.3818013 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 318 cuvinte
Subiecte conexe
ANCRe: Adaptive Neural Connection Reassignment for Efficient Depth Scaling
arXiv · 6 oct. 2026
SPECOLA: A Tool for Specification Driven Robot Evaluation
arXiv · 27 oct. 2026
Quantum Keys Travel 18 Kilometers Across Hybrid Air-and-Fiber Network
The Quantum Insider · 26 aug. 2026
LARC: Low-Rank Adaptive Residual Connections for Learning in Frozen Models
arXiv · 1 oct. 2026