ZeroDiff: Zero-Shot Time Series Reconstruction via Informed-Prior Diffusion
ZeroDiff:基于先验信息扩散的零样本时间序列重建
机构 * University of Pittsburgh(匹兹堡大学) ; Oak Ridge National Laboratory(橡树岭国家实验室)
AI总结 针对目标观测稀缺的零样本时间序列重建问题,提出ZeroDiff方法,利用外生变量构建先验并通过扩散校准误差,在真实数据集上显著优于现有方法。
Comments ICML 2026. Code: this https URL (ICML2026" target="_blank" rel="noopener">https://github.com/YingdaFan/ZeroDiff-ICML2026)