message-passing
Message-passing refers to the process by which nodes in graphical models or neural networks communicate information to one another. This technique is central to many algorithms in probabilistic programming and graph-based deep learning.
- A Signed Graph Approach to Understanding and Mitigating Oversmoothing
- Enhancing Graph Classification Robustness with Singular Pooling
- Generative Graph Pattern Machine
- Graph Persistence goes Spectral
- Graph-KV: Breaking Sequence via Injecting Structural Biases into Large Language Models
- Higher-Order Learning with Graph Neural Networks via Hypergraph Encodings
- On Local Limits of Sparse Random Graphs: Color Convergence and the Refined Configuration Model
- On Vanishing Gradients, Over-Smoothing, and Over-Squashing in GNNs: Bridging Recurrent and Graph Learning
- The Underappreciated Power of Vision Models for Graph Structural Understanding