2020/10/08 by Leonardo Cotta, Carlos H. C. Teixeira, Cotta, Leonardo +5
Computer Science · Materials Science · Physics and Astronomy · #Advanced Graph Neural Networks #Artificial Intelligence (cs.AI) #Complex Network Analysis Techniques #FOS: Computer and information sciences #Graph Theory and Algorithms #Machine Learning (cs.LG) #Machine Learning in Materials Science
paper · pdf · doi:10.48550/arxiv.2010.04259
openalex publication_date 2020/10/08 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28
Existing Graph Neural Network (GNN) methods that learn inductive unsupervised\ngraph representations focus on learning node and edge representations by\npredicting observed edges in the graph. Although such approaches have shown\nadvances in downstream node classification tasks, they are ineffective in\njointly representing larger k-node sets, k>2. We propose MHM-GNN, an\ninductive unsupervised graph representation approach that combines joint\nk-node representations with energy-based models (hypergraph Markov networks)\nand GNNs. To address the intractability of the loss that arises from this\ncombination, we endow our optimization with a loss upper bound using a\nfinite-sample unbiased Markov Chain Monte Carlo estimator. Our experiments show\nthat the unsupervised MHM-GNN representations of MHM-GNN produce better\nunsupervised representations than existing approaches from the literature.\n