A multi-stage GNN with hierarchical pooling and unpooling predicts natural-convection temperature fields more accurately and efficiently than a single-scale MeshGraphNets baseline on a new 2D cavity dataset.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Multi-Stage Graph Neural Networks for Data-Driven Prediction of Natural Convection in Enclosed Cavities
A multi-stage GNN with hierarchical pooling and unpooling predicts natural-convection temperature fields more accurately and efficiently than a single-scale MeshGraphNets baseline on a new 2D cavity dataset.