Demin Yu (于德民)

Ph.D. Candidate

School of Computer Science and Technology
Harbin Institute of Technology
HIT Campus of University Town of Shenzhen, Guangdong, China

E-mail: deminyu98@gmail.com

Google Scholar ResearchGate GitHub

I am actively seeking postdoctoral opportunities in research groups working on AI for meteorology, weather/climate intelligence, and related spatio-temporal modelling problems. I would be very happy to discuss potential collaborations with groups sharing similar research interests.

Biography

I am a Ph.D. candidate in Computer Science and Technology at Harbin Institute of Technology, Shenzhen, advised by Prof. Xutao Li and Prof. Yunming Ye. I received my M.E. degree from Harbin Institute of Technology, Harbin, in 2022, and my B.E. degree from Harbin Institute of Technology, Weihai, in 2020.

My research lies at the intersection of artificial intelligence and meteorology, with broad interests in AI for meteorology, computer vision, generative modelling, and spatio-temporal learning. Methodologically, I focus on building learning frameworks that combine physical structure, probabilistic generation, and temporal dynamics to model atmospheric evolution, including precipitation nowcasting, satellite-to-radar inversion, and multi-source weather forecasting.

Education

Ph.D. (2026.09 expected), School of Computer Science and Technology, Harbin Institute of Technology, Shenzhen, China.
M.E. (2022.06), School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China.
B.E. (2020.06), School of Computer Science and Technology, Harbin Institute of Technology, Weihai, China.

Publications [Google Scholar]

Preprints and Under Review

image pending

AlignCast: Align-Then-Cast Radar Nowcasting with Time-Asynchronous Satellite Guidance

Demin Yu, et al.

Under review, 2026

AlignCast first learns radar-compatible satellite representations, then uses time-aware cross-attention to handle temporal asynchrony between radar and satellite observations.

image pending

MPT: Motion-Conditioned Posterior Transport for Sequence Satellite-to-Radar Inversion

Demin Yu, et al.

Under review, 2026

MPT uses storm motion inferred from temporal satellite geometry to constrain plausible radar evolution through a deterministic anchor and motion-conditioned rectified flow.

First-Author Publications

PiMMNet: Introducing Multi-Modal Precipitation Nowcasting via a Physics-informed Perspective

Demin Yu, Wenchuan Du, Kenghong Lin, Xutao Li, Yunming Ye, Chuyao Luo, Xunlai Chen.

ACM International Conference on Multimedia (ACM MM), 2025. [CORE A*]

PiMMNet models multi-modal meteorological evolution as advection-diffusion under a shared latent velocity field, treating radar and satellite as observations of one evolving physical state.

[DOI]

MMCast: Integrating Multi-Source Data for Long Sequence Precipitation Forecasting

Demin Yu, Wenzhi Feng, Kenghong Lin, Xutao Li, Yunming Ye, Chuyao Luo, Wenchuan Du.

AAAI Conference on Artificial Intelligence, 2025. [CORE A*]

MMCast combines radar with broader satellite context for long-sequence precipitation forecasting and extends useful forecasts toward multi-hour horizons.

[DOI]

DiffCast: A Unified Framework via Residual Diffusion for Precipitation Nowcasting

Demin Yu, Xutao Li, Yunming Ye, Baoquan Zhang, Chuyao Luo, Kuai Dai, Rui Wang, Xunlai Chen.

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024. [CORE A*]

DiffCast decomposes precipitation into predictable large-scale motion and stochastic small-scale intensity change, recovering sharper echoes while preserving storm positions.

[arXiv]

Co-Authored Publications

	Four-hour thunderstorm nowcasting using deep diffusion models of satellite Kuai Dai, Xutao Li, Junying Fang, Yunming Ye, Demin Yu, Hui Su, Di Xian, Danyu Qin, Jingsong Wang. Proceedings of the National Academy of Sciences (PNAS), 2025 [arXiv]
	AlphaPre: Amplitude-Phase Disentanglement Model for Precipitation Nowcasting Kenghong Lin, Baoquan Zhang, Demin Yu, Wenzhi Feng, Shidong Chen, Feifan Gao, Xutao Li, Yunming Ye. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025. [CORE A*] [DOI]
	PercepCast: Perceptually Constrained Precipitation Nowcasting Model Wenzhi Feng, Xutao Li, Zhe Wu, Kenghong Lin, Demin Yu, Yunming Ye, Yaowei Wang. International Conference on Machine Learning (ICML), 2025. [CORE A*]
	LMcast: A Pretrained Language Model Guided Long-Term Memory Transformer for Precipitation Nowcasting Feifan Gao, Chuyao Luo, Guangbo Deng, Xutao Li, Baoquan Zhang, Demin Yu, Yunming Ye. Neural Networks, 2025 [DOI]
	MetaDiff: Meta-Learning with Conditional Diffusion for Few-Shot Learning Baoquan Zhang, Chuyao Luo, Demin Yu, Xutao Li, Huiwei Lin, Yunming Ye, Bowen Zhang. AAAI Conference on Artificial Intelligence, 2023. [CORE A*] [arXiv]

Professional Service

Conference reviewer: CVPR, NeurIPS, ECCV.

Teaching

Teaching Assistant, Compilation Principles, Spring 2023.
Teaching Assistant, Data Mining, Fall 2024.