aiqbench analysis
This research reveals that standard message passing in Graph Contrastive Learning can trivialize the learning signal from positive samples, a flaw addressed by the new SPGCL method. By filtering for high Dirichlet energy features to capture effective signals while using low-energy features for sampling, the approach restores the model's ability to learn intrinsic graph semantics. The technique offers a more robust pre-training paradigm for graph neural networks by decoupling feature propagation from positive sample maximization.
Revisiting Positive Samples in Graph Contrastive Learning: From the Perspective of Message Passing
arXiv:2606.10284v1 Announce Type: new Abstract: Graph Contrastive Learning (GCL), which trains graph encoders by maximizing similarity between positive samples and minimizing it between negative ones, has emerged as a mainstream graph pre-training paradigm. It is widely recognized that positive samples are essential in GCLs. Ideally, maximizing the similarity of positive samples enables graph encoders to capture intrinsic semantic and patterns of graph data. However, we discover an interesting phenomenon: GCLs can achieve competitive performance even without positive samples. This motivates us to revisit the fundamental mechanism of positive samples in GCLs. From the perspective of Dirichlet energy, we theoretically finds that message passing, a key mechanism in graph encoders, trivializes the maximization of positive samples, preventing GCLs from effectively learning from positive samples. To address this, we propose SPGCL to mitigate the trivialization caused by message passing and restore the learning efficacy of positive samples. Specifically, we find that high Dirichlet energy features help positive samples provide effective learning signals while low Dirichlet energy features contribute little to positive learning signal but is useful for positive sampling. Based on this, SPGCL propagates only high Dirichlet energy features and uses low energy features to construct a probability matrix for reliable positive sampling. Extensive experiments demonstrate the effectiveness of SPGCL.
Abstract:Graph Contrastive Learning (GCL), which trains graph encoders by maximizing similarity between positive samples and minimizing it between negative ones, has emerged as a mainstream graph pre-training paradigm. It is widely recognized that positive samples are essential in GCLs. Ideally, maximizing the similarity of positive samples enables graph encoders to capture intrinsic semantic and patterns of graph data. However, we discover an interesting phenomenon: GCLs can achieve competitive performance even without positive samples. This motivates us to revisit the fundamental mechanism of positive samples in GCLs. From the perspective of Dirichlet energy, we theoretically finds that message passing, a key mechanism in graph encoders, trivializes the maximization of positive samples, preventing GCLs from effectively learning from positive samples. To address this, we propose SPGCL to mitigate the trivialization caused by message passing and restore the learning efficacy of positive samples. Specifically, we find that high Dirichlet energy features help positive samples provide effective learning signals while low Dirichlet energy features contribute little to positive learning signal but is useful for positive sampling. Based on this, SPGCL propagates only high Dirichlet energy features and uses low energy features to construct a probability matrix for reliable positive sampling. Extensive experiments demonstrate the effectiveness of SPGCL.
Machine Learning (cs.LG)
arXiv:2606.10284 [cs.LG]
(or arXiv:2606.10284v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.10284 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.
Full text under the publisher's open licence · 297 words
Related coverage
Examining AI Swarm Efficiency with the AI Futures Model
LessWrong · Oct 19, 2026
Revisiting Spectral Representations in Generative Diffusion Models
arXiv · Sep 9, 2026
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026
OpenRouter: from Seed to Stripe — with OpenRouter’s Alex Atallah & AMP’s Anjney Midha
Latent Space · Sep 25, 2026