I am Xiangyu Chen, currently a Ph.D. candidate at Tsinghua-Berkeley Shenzhen Institute (TBSI), focusing on Data Science and Machine Learning. I am also a visiting scholar at UC Berkeley, working on AI for Science projects, specifically on graphical battery capability prediction.

My research interests include:

• Neural Decoder Design and Optimization
• Transfer Learning and Meta Learning
• Graph Learning and Model Recommendation
• Weakly Supervised Learning

Education

University of California, Berkeley (2023.07 - 2024.03)
Visiting Scholar in AI for Science
Advisor: Prof. Khalid M. Mosalam
Research: Graphical battery capability prediction

Tsinghua-Berkeley Shenzhen Institute (2021.06 - 2024.09)
Ph.D. Candidate in Data Science
Advisor: Prof. Yang Li
Research: Neural decoder, transfer learning and meta learning

Tsinghua Shenzhen International Graduate School (2018.09 - 2021.07)
M.S. in Data Science
Advisor: Prof. Yong Jiang
GPA: 3.84/4.0

Wuhan Institute of Technology (2013.09 - 2016.06)
B.S. in Process Equipment and Control Engineering
National Encouragement Scholarship

Research Highlights

My research work focuses on machine learning algorithm design and optimization. Key publications include:

1. Neural Decoder (ICML 2021, Oral < 3%)
   • Developed Cyclically Equivariant Neural Decoders for Cyclic Codes
   • Implemented shift-invariant structure achieving near Maximum Likelihood decoder performance
   • Project Page
2. List Decoder (ISIT 2022)
   • Enhanced the list decoding version of the Cyclically Equivariant Neural Decoder
3. Affine Decoder (Journal of the Franklin Institute 2023)
   • Designed neural decoders with permutation invariant structure
4. Graph Learning (Submitted to CIKM 2024)
   • Proposed a graph learning-based approach for model transferability prediction
   • Paper Preview
5. Weakly Supervised Learning (CVPR 2020)
   • Introduced JoCoR, a robust learning paradigm for noisy label scenarios
   • Achieved significant impact with 477 citations

Technical Skills

• Programming Languages: Python, C/C++, C#, HTML, JavaScript
• ML Frameworks: TensorFlow, Keras, PyTorch
• Web Development: Flask, Bootstrap