Incorporating Domain Differential Equations into Graph Convolutional Networks to Lower Generalization Discrepancy

arXiv:2404.01217v1 Announce Type: new
Abstract: Ensuring both accuracy and robustness in time series prediction is critical to many applications, ranging from urban planning to pandemic management. With sufficient training data where all spatiotemporal patterns are well-represented, existing deep-learning models can make reasonably accurate predictions. However, existing methods fail when the training data are drawn from different circumstances (e.g., traffic patterns on regular days) compared to test data (e.g., traffic patterns after a natural disaster). Such challenges are usually classified under …

abstract accuracy applications arxiv cs.ai cs.lg data differential domain graph however management networks pandemic patterns planning prediction predictions robustness series time series training training data type urban urban planning