Haowen Wang | 王皓雯

I'm Haowen Wang, a fourth-year undergraduate student at School of Software, Tsinghua University majoring in Software Engineering.

I'm honored to work with Prof. Mingsheng Long, Prof. Yue Gao, and Prof. Xibin Zhao at Tsinghua University. In addition, I am fortunate to be a research visitor at University of California, San Diego, working with Prof. Zhuowen Tu this summer.

My research interests lie in AI4Science, Computer Vision and generative AI. My research goal is to help machines see and understand this world through scientific and interpretable ways.

Email  /  CV  /  Google Scholar  /  Github

profile photo

Publications (* indicates equal contribution)

Transolver: A Fast Transformer Solver for PDEs on General Geometries
Haixu Wu, Huakun Luo, Haowen Wang, Jianming Wang, Mingsheng Long
ICML 2024 (Spotlight Paper)
PDF | Code | Slides

We designed a physics-informed model to solve partial differential equations (PDEs), learning intrinsic physical states for better performance.

Hypergraph-Guided Disentangled Spectrum Transformer Networks for Near-Infrared Facial Expression Recognition
Bingjun Luo, Haowen Wang, Jinpeng Wang, Junjie Zhu, Xibin Zhao, Yue Gao
AAAI 2024
PDF

We proposed a hypergraph-guided disentangled spectrum transformer network for near-infrared facial expression recognition.

Unisolver: PDE-Conditional Transformers Are Universal PDE Solvers
Hang Zhou*, Yuezhou Ma*, Haixu Wu, Haowen Wang, Mingsheng Long
ICLR 2025 FM-Wild Workshop
PDF

We presents the Universal PDE solver (Unisolver) capable of solving a wide scope of PDEs by leveraging a Transformer pre-trained on diverse data and conditioned on diverse PDEs.

Selected Awards

  • 2024: National Scholarship, Ministry of Education (top scholarship in China, 0.2% domestically, 国家奖学金)
  • 2023: Comprehensive Excellence Scholarship, Tsinghua University (top 1%)
  • 2022: Comprehensive Excellence Scholarship, Tsinghua University (top 1%)
  • Misc

  • Badminton: I love playing badminton🏸️ ! Can't live without it 🥰
  • Swimming: 🏊🏻🩱🐟🏖️

  • The source code is stolen from Jon Barron. Thanks for his sharing! 🙏🏻