About Me
I am currently a Research Assistant at The Hong Kong University of Science and Technology (Guangzhou) (HKUST(GZ)), working with Prof. Kaishun Wu since July 2026. I received my M.S. (2026) in Computer Technology from Shenzhen University, supervised by Prof. Haodi Zhang, and my B.S. (2023) in Information and Computational Science from Guangdong University of Technology.
My research lies at the intersection of AI for Science and intelligent decision-making, with applications in agriculture and climate. I combine physical modeling, domain knowledge, and reinforcement learning to study how AI can understand environmental processes and support decisions under practical constraints. My recent work includes neural operators for meteorological downscaling and learning-based optimization for agricultural mitigation and livestock planning.
Research Interests
- Physics-Informed Environment Modeling — Combining physical laws and neural operators to model environmental processes across spatial scales.
- Knowledge-Integrated Spatial Decision-Making — Integrating domain knowledge and reinforcement learning for spatial planning under resource and environmental constraints.
- Knowledge-Aware Scientific Agents — Connecting environmental models, scientific knowledge, and agent reasoning to support long-term planning.
My long-term goal is to build reliable scientific agents that turn an understanding of the physical world into informed action, helping address challenges in sustainable agriculture and climate adaptation.
🔥 News
- 2026.7: I joined HKUST(GZ) as a Research Assistant, supervised by Prof. Kaishun Wu.
- 2026.4.28: MSPNO was accepted by IJCAI 2026. It learns high-resolution meteorological fields with spherical, multi-scale, and physics-informed neural operators.
📖 Education
- 2023 - 2026 | M.S. in Computer Technology
Shenzhen University, Shenzhen, China
Supervised by Prof. Haodi Zhang.
Research on physics-informed operator learning and knowledge-enhanced decision optimization. - 2019 - 2023 | B.S. in Information and Computational Science
Guangdong University of Technology, Guangzhou, China
Recommended for Admission.
💻 Research
A spherical, multi-scale, and physics-informed operator learning framework for high-resolution climate field modeling. The project combines spherical Laplacian decomposition, localized spherical integral operators, and physics-informed constraints to improve geometric continuity and physical consistency.
A county-level technology portfolio optimization pipeline covering 2,901 counties, 105 mitigation technologies, 18 carbon and nitrogen indicators, 4,538 publications, and 22,635 observations. PTO combines expert-guided initialization and curriculum learning to reduce emissions, cost, and implementation complexity.
A constrained actor-critic planning framework for livestock redistribution under agronomic, environmental, and social constraints. The framework uses PPO, action masking, hierarchical migration/intervention stages, and DRL+LP optimization. Scenario results report 63% nitrogen surplus reduction, 48% ammonia-emission-intensity reduction, and 71% pollution-exposure reduction.
📝 Selected Publications
Spherical Physics-informed Neural Operator with Multi-scale Coupling for Meteorological Downscaling
[PDF]Interpretable Pulmonary Disease Diagnosis with Graph Neural Network and Counterfactual Explanations
🏆 Honors & Awards
- 2024–2025: Outstanding Student Scholarship, Second Prize, Shenzhen University.
- 2023: Grand Prize Student Scholarship, Shenzhen University.
- 2019–2023: Outstanding Student Scholarship, Third Prize, Guangdong University of Technology.
📬 Contact
I am open to research discussions on AI4Science, physics-informed operator learning, constrained reinforcement learning, and knowledge-aware scientific agents.
- Email: yaniszz085@gmail.com
- CV: Yiqiang-Ye-CV-EN.pdf