analiza aiqbench
This research establishes a rigorous protocol for cross-task transfer in Graph Neural Networks (GNNs), demonstrating that supervision from node classification can reliably enhance link prediction on homophilic graphs, whereas the reverse direction is often unstable. The authors introduce the CoTask Score to quantify joint utility and show that simple dataset statistics like homophily effectively predict whether shared encoders will succeed or suffer negative transfer. These findings provide critical guidance for optimizing multi-task learning strategies in graph-based machine learning systems without relying on flawed evaluation splits.
Same Graph Cross-Task Transfer in GNNs: Protocols and Predictors
arXiv:2607.28525v1 Announce Type: new Abstract: Many real-world graphs support multiple predictive tasks over the same underlying structure, creating an opportunity to reuse supervision across node classification (NC) and link prediction (LP). However, existing evaluations often rely on incompatible splits, observed-graph assumptions, and negative sampling rules, making conclusions about same-graph cross-task transfer unreliable. We formalize same-graph NC-LP transfer and propose a leakage-free protocol that fixes node and edge splits, uses a shared message-passing graph that excludes evaluated edges, and employs fixed negatives for LP. Across three backbones (GCN, GraphSAGE, GPS), we find that transfer is strongly directional and predictable: NC $\to$ LP is consistently beneficial on homophilic graphs, while LP $\to$ NC is fragile and can even degrade accuracy under naive representation reuse. LP $\to$ NC becomes reliably positive mainly in a structure-dominant regime where LP is easy but NC is unsaturated, suggesting that LP acts as structural pretraining. Finally, we introduce the CoTask Score (CTS) to summarize joint NC+LP utility when a shared encoder must serve both tasks, and show that simple dataset statistics, especially homophily, can guide mechanism choice and help avoid negative transfer.
Abstract:Many real-world graphs support multiple predictive tasks over the same underlying structure, creating an opportunity to reuse supervision across node classification (NC) and link prediction (LP). However, existing evaluations often rely on incompatible splits, observed-graph assumptions, and negative sampling rules, making conclusions about same-graph cross-task transfer unreliable. We formalize same-graph NC-LP transfer and propose a leakage-free protocol that fixes node and edge splits, uses a shared message-passing graph that excludes evaluated edges, and employs fixed negatives for LP. Across three backbones (GCN, GraphSAGE, GPS), we find that transfer is strongly directional and predictable: NC $\to$ LP is consistently beneficial on homophilic graphs, while LP $\to$ NC is fragile and can even degrade accuracy under naive representation reuse. LP $\to$ NC becomes reliably positive mainly in a structure-dominant regime where LP is easy but NC is unsaturated, suggesting that LP acts as structural pretraining. Finally, we introduce the CoTask Score (CTS) to summarize joint NC+LP utility when a shared encoder must serve both tasks, and show that simple dataset statistics, especially homophily, can guide mechanism choice and help avoid negative transfer.
Machine Learning (cs.LG); Social and Information Networks (cs.SI)
arXiv:2607.28525 [cs.LG]
(or arXiv:2607.28525v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.28525 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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 · 279 cuvinte
Subiecte conexe
ProMeta: Few-shot PROTAC-targeted degradation prediction across E3 ligases
arXiv · 10 sept. 2026
Benchmarking Sheaf Neural Networks for Inductive Tasks
arXiv · 4 aug. 2026
Chimaera: A Mixture-of-Graph-Experts Architecture for Cross-Task and Cross-Dataset Graph Learning
arXiv · 9 sept. 2026
LLM as GNN: Graph Vocabulary Learning for Text-Attributed Graph Foundation Models
arXiv · 4 sept. 2026