
STD-PLM, a unified framework based on pre-trained language models, can effectively understand both spatial and temporal properties of spatial-temporal data, enabling competitive performance in forecasting and imputation tasks while exhibiting strong few-shot and zero-shot learning capabilities.


coremsg

unified-pre-trained-spatial-temporal-model-for-forecasting-and-imputation-with-few-shot-and-zero-shot-capabilities


Unified Pre-trained Spatial-Temporal Model for Forecasting and Imputation with Few-Shot and Zero-Shot Capabilities
