Scholar
liu ziyin
Google Scholar ID: P5fPJ1oAAAAJ
MIT, NTT Research
Citations & Impact
Citations
2,060
H-index
18
i10-index
23
Publications
20
Co-authors
14
list available
Contact
Publications
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
- Selected Publications: 'Topological Invariance and Breakdown in Learning' (arXiv), 'A universal compression theory: Lottery ticket hypothesis and superpolynomial scaling laws' (arXiv), 'Neural Thermodynamics I: Entropic Forces in Deep and Universal Representation Learning' (NeurIPS 2025), 'Heterosynaptic Circuits Are Universal Gradient Machines' (Preprint 2025). Conference Papers: ICLR 2022, ICML 2024, NeurIPS 2025, etc. Served as an Area Chair for NeurIPS and ICLR.
Research Experience
- Works with Prof. Isaac Chuang at MIT and collaborates with Prof. Tomaso Poggio in the BCS department. Focuses on the theoretical foundation of deep learning. Previously, pursued a PhD in Physics at the University of Tokyo.
Education
- PhD in Physics from the University of Tokyo, supervised by Prof. Masahito Ueda; Bachelor's degree in Physics and Mathematics from Carnegie Mellon University
Background
- Research Interests: Scientific principles of artificial intelligence, theories of deep learning, universal laws of AI, grand unified theory of intelligence. Brief Introduction: A researcher at MIT and NTT Research, focusing on the intersection of mathematics, physics, neuroscience, and artificial intelligence.
Miscellany
- Personal interests include art, literature, and philosophy. Enjoys playing Go.
