Liu Ziyin
Office: Room 26-209, MIT
I am a research scientist at MIT. My research lies at the mysterious-and-fantastic union and intersection of mathematics, physics, neuroscience and artificial intelligence. At MIT, I work with Isaac Chuang and Tomaso Poggio. My research focus is on the theoretical foundation of deep learning. Prior to coming to MIT, I received my PhD in physics at the University of Tokyo under the supervision of Masahito Ueda. I received a Bachelor's degree in physics and mathematics at Carnegie Mellon University. I also serve as an Area Chair for NeurIPS, ICML, and ICLR. Personally, I am interested in art, literature, and philosophy. I also play Go. If you have questions or want to collaborate, or just want to say hi, feel free to send an email.
About my name: My first name is Ziyin and last name is Liu, though on publications I stick to the the eastern convention of [last name] [first name]. So please feel free to call me Ziyin or Dr. Liu, but when adding my name to publications write "Liu Ziyin."
Talks I gave recently:
Universal Phenomena, Irreversibility, and Thermodynamics in Deep Representation Learning
How does physics help understand deep learning?
Doctor thesis: Symmetry breaking in deep learning (深層学習に於ける対称性の破れ, 2023).
Master thesis: Mean-field learning dynamics of deep neural networks (2020).
Research Interest
I am particularly interested in identifying scientific principles of both artificial and biological intelligence (what is a principle?), and I think tools and intuitions from other scientific fields can help. On the AI side, I work on:
- Theories of deep learning
- Universal laws of AI (!!!)
- Grand unified theory of intelligence (!!!!!)
On the neuroscience side, I work on
- Connectomics
- Principles of circuit organization in the brain
For my selected works, see below, and for my full list of publications and preprints, see publications.
Tutorial / Notes
Proof of a perfect platonic representation hypothesis (2025)
Parameter Symmetry Potentially Unifies Deep Learning Theory (2025)
Selected Work
(* denotes equal contribution or corresponding author)
-
Connectome-based biophysical modeling of a figure-ground discrimination circuit [bioRxiv]
Liu Ziyin, Yifei Lu, Alexander Borst, Sven Dorkenwald, Tomaso Poggio
Preprint 2026
-
A universal compression theory for lottery ticket hypothesis and neural scaling laws [arXiv]
Hong-Yi Wang, Di Luo, Tomaso Poggio, Isaac L. Chuang, Liu Ziyin*
ICLR 2026
-
Topological Invariance and Breakdown in Learning Dynamics [arXiv]
Yongyi Yang, Tomaso Poggio, Isaac Chuang, Liu Ziyin*
NeurIPS 2026
-
How Biological Synapses Self-Assemble Gradient Learning [paper] [arXiv]
Qianli Liao*, Liu Ziyin*, Yulu Gan*, Brian Cheung, Mark T. Harnett, Tomaso Poggio
PNAS 2026
-
Neural Thermodynamics: Entropic Forces in Deep and Universal Representation Learning [arXiv]
Liu Ziyin*, Yizhou Xu*, Isaac Chuang
NeurIPS 2025
-
Heterosynaptic Circuits Are Universal Gradient Machines [arXiv]
Liu Ziyin, Isaac Chuang, Tomaso Poggio
Preprint 2025
-
Formation of Representations in Neural Networks [paper]
Liu Ziyin, Isaac Chuang, Tomer Galanti, Tomaso Poggio
ICLR 2025 (spotlight: 5% of all submissions)
-
Parameter Symmetry and Noise Equilibrium of Stochastic Gradient Descent [arXiv]
Liu Ziyin, Mingze Wang, Hongchao Li, Lei Wu
NeurIPS 2024
-
Symmetry Induces Structure and Constraint of Learning [arXiv]
Liu Ziyin
ICML 2024
-
Sparsity by Redundancy: Solving L1 with SGD [arXiv]
Liu Ziyin*, Zihao Wang*
ICML 2023
-
Neural Networks Fail to Learn Periodic Functions and How to Fix It [arXiv]
Liu Ziyin, Tilman Hartwig, Masahito Ueda
NeurIPS 2020
This page has been accessed
times since July 07, 2018.