Official implementation of MCSTL — Mask- and Contrast-Enhanced Spatio-Temporal Learning for Urban Flow Prediction (CIKM 2023). MCSTL pre-trains an urban flow model with a mask-reconstruction task across space and time and a graph-based contrastive task that weights regions by inter-regional attention, then fine-tunes for flow prediction.
📄 Paper: https://doi.org/10.1145/3583780.3614958 · 🌐 Project page with abstract, FAQ and BibTeX: https://codezx6.github.io/papers/mcstl.html · 👤 Author: Xu Zhang
@inproceedings{zhang2023mcstl,
title = {Mask- and Contrast-Enhanced Spatio-Temporal Learning for Urban Flow Prediction},
author = {Zhang, Xu and Gong, Yongshun and Zhang, Xinxin and Wu, Xiaoming and Zhang, Chengqi and Dong, Xiangjun},
booktitle = {Proceedings of the 32nd ACM International Conference on Information and Knowledge Management (CIKM '23)},
year = {2023},
pages = {3298--3307},
publisher = {ACM},
doi = {10.1145/3583780.3614958},
url = {https://doi.org/10.1145/3583780.3614958}
}