Research Assistant, HKUST (GZ)
Research Intern, 2012 Laboratories, HuaweiJunjie Qiu is an incoming PhD student in Data Science and Analytics at the Hong Kong University of Science and Technology (Guangzhou). He is currently a Research Assistant in the Intelligent Transportation Thrust, working under the supervision of Prof. Yuxuan Liang. Previously, he was a Research Intern at Huawei’s 2012 Laboratories.
Warning
Problem: The current name of your GitHub Pages repository ("Solution: Please consider renaming the repository to "
http://".
However, if the current repository name is intended, you can ignore this message by removing "{% include widgets/debug_repo_name.html %}" in index.html.
Action required
Problem: The current root path of this site is "baseurl ("_config.yml.
Solution: Please set the
baseurl in _config.yml to "Education
-
Hong Kong University of Science and Technology (GZ)Ph.D. in Data Science and AnalyticsUpcoming
-
Southern University of Science and TechnologyB.S. in Data Science and Big Data TechnologySep. 2021 - Jul. 2025
Work Experience
-
2012 Laboratories, HuaweiResearch Intern, Spatio-Temporal Foundation ModelMar. 2025 - Present
News
Selected Publications (view all )
Learning to Factorize Spatio-Temporal Foundation Models
Siru Zhong, Junjie Qiu, Yangyu Wu, Xingchen Zou, Bin Yang, Chenjuan Guo, Hao Xu, Yuxuan Liang# (# corresponding author)
Neural Information Processing Systems (NeurIPS) 2025 Spotlight
Spatio-Temporal (ST) Foundation Models (STFMs) promise cross-dataset generalization, yet joint ST pretraining is computationally costly and struggles with domain-specific spatial correlations. To address this, we propose FactoST, a factorized STFM that decouples universal temporal pretraining from ST adaptation. The first stage trains a space-agnostic backbone via multi-task learning to capture multi-frequency, cross-domain temporal patterns at low cost. The second stage attaches a lightweight adapter that rapidly adapts the backbone to specific ST domains via metadata fusion, interaction pruning, domain alignment, and memory replay. Extensive forecasting experiments show that in few-shot settings, FactoST reduces MAE by up to 46.4% versus UniST, uses 46.2% fewer parameters, achieves 68% faster inference than OpenCity, and remains competitive with expert models. This factorized view offers a practical, scalable path toward truly universal STFMs.
Learning to Factorize Spatio-Temporal Foundation Models
Siru Zhong, Junjie Qiu, Yangyu Wu, Xingchen Zou, Bin Yang, Chenjuan Guo, Hao Xu, Yuxuan Liang# (# corresponding author)
Neural Information Processing Systems (NeurIPS) 2025 Spotlight
Spatio-Temporal (ST) Foundation Models (STFMs) promise cross-dataset generalization, yet joint ST pretraining is computationally costly and struggles with domain-specific spatial correlations. To address this, we propose FactoST, a factorized STFM that decouples universal temporal pretraining from ST adaptation. The first stage trains a space-agnostic backbone via multi-task learning to capture multi-frequency, cross-domain temporal patterns at low cost. The second stage attaches a lightweight adapter that rapidly adapts the backbone to specific ST domains via metadata fusion, interaction pruning, domain alignment, and memory replay. Extensive forecasting experiments show that in few-shot settings, FactoST reduces MAE by up to 46.4% versus UniST, uses 46.2% fewer parameters, achieves 68% faster inference than OpenCity, and remains competitive with expert models. This factorized view offers a practical, scalable path toward truly universal STFMs.
All publications
Honors & Awards
-
Top Ten Graduate Award, Shude Residential College, SUSTechJune. 2025
-
First Prize (10/300+) & Group Competitions Prize, ASC24Apr. 2024
Service
-
Student Member, IEEENov. 2024 - Present