Scholar
Diyuan Wu
Google Scholar ID: MztcJLMAAAAJ
IST Austria
Citations & Impact
Citations
21
H-index
2
i10-index
1
Publications
5
Co-authors
9
list available
Contact
Publications
Resume (English only)
Academic Achievements
- Publications:
- - "Attention with Trained Embeddings Provably Selects Important Tokens" (with Alex, Samet, Marco), accepted to NeurIPS 2025.
- - "Neural Collapse Beyond the Unconstrained Features Model: Landscape, Dynamics, and Generalization in the Mean-Field Regime" (with Marco), accepted as a spotlight paper at ICML 2025.
Research Experience
- Currently a PhD student at the Institute of Science and Technology Austria (ISTA) working on machine learning theory.
Education
- PhD: Institute of Science and Technology Austria (ISTA), started in fall 2022, supervised by Professor Marco Mondelli; Master's degree: EPFL, Lausanne, Switzerland, graduated in 2022; Bachelor's degree: UESTC (University of Electronic Science and Technology of China), Chengdu, China, graduated in 2019.
Background
- Research interests: Machine learning theory, particularly on high-dimensional data/model problems. Specific research areas include feature learning and weak-to-strong generalization.
Miscellany
- Contact: Email diyuan.wu@ist.ac.at
