Основные понятия
C2TSD is a novel framework that integrates temporal disentanglement and contrastive learning for superior spatiotemporal imputation.
Аннотация
The article introduces C2TSD, a conditional diffusion framework for spatiotemporal imputation. It addresses challenges in generating stable results by incorporating disentangled temporal representations and contrastive learning. The model outperforms state-of-the-art baselines on real-world datasets from meteorology and transportation fields.
JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021
Spatiotemporal data analysis is crucial in various domains.
Traditional imputation methods face challenges with incomplete data.
Generative models like diffusion models offer potential solutions.
C2TSD framework incorporates disentangled temporal representations and contrastive learning.
Extensive experiments show superior performance compared to baselines.
Data Extraction:
"Our contributions are summarized as follows" - Extensive experiments on three real-world datasets demonstrate the superior performance of our approach compared to a number of state-of-the-art baselines.
Quotations:
"We propose C2TSD, a contrastive diffusion framework for spatiotemporal imputation."
"Our contributions are summarized as follows."
Статистика
Our contributions are summarized as follows - Extensive experiments on three real-world datasets demonstrate the superior performance of our approach compared to a number of state-of-the-art baselines.
Цитаты
We propose C2TSD, a contrastive diffusion framework for spatiotemporal imputation.
Our contributions are summarized as follows.