2021/06/30 by Charilaos Mylonas, Mylonas, Charilaos, Imad Abdallah +3 · 1 citation
Computer Science · #Advanced Graph Neural Networks #Bayesian Modeling and Causal Inference #Computational Engineering #FOS: Computer and information sciences #Finance #Gaussian Processes and Bayesian Inference #Machine Learning (cs.LG) #Machine Learning (stat.ML) #and Science (cs.CE)
paper · pdf · doi:10.48550/arxiv.2106.16049
openalex publication_date 2021/06/30 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Graph Networks (GNs) enable the fusion of prior knowledge and relational\nreasoning with flexible function approximations. In this work, a general\nGN-based model is proposed which takes full advantage of the relational\nmodeling capabilities of GNs and extends these to probabilistic modeling with\nVariational Bayes (VB). To that end, we combine complementary pre-existing\napproaches on VB for graph data and propose an approach that relies on\ngraph-structured latent and conditioning variables. It is demonstrated that\nNeural Processes can also be viewed through the lens of the proposed model. We\nshow applications on the problem of structured probability density modeling for\nsimulated and real wind farm monitoring data, as well as on the meta-learning\nof simulated Gaussian Process data. We release the source code, along with the\nsimulated datasets.\n