A Multitask Learning View on the Earth System Model Ensemble

作者：Goncalves Andre R^*; Von Zuben Fernando J; Banerjee Arindam

来源：Computing in Science & Engineering, 2015, 17(6): 35-42.

DOI：10.1109/mcse.2015.105

Earth system models (ESMs) are the primary mechanisms for obtaining projections under different climate change scenarios. Researchers use ensembles of climate models to gain better accuracy and reduce uncertainty. A multitask learning-based method can build ESM ensembles for all regions jointly to improve predictions for individual ones.

• 出版日期2015-12