Research

Physics

Topological Invariance and Breakdown in Learning Dynamics
  1. Learning the identity: a case study of how SGD selects among functional decompositionsPreprint 2026
  2. Width-Independent Compressibility of Deep Neural NetworksPreprint 2026
  3. Neural Quadratic Forms: A Unified Minimal Model for Sudden Learning and Scaling LawsPreprint 2026
  4. Thermodynamic Irreversibility of Training AlgorithmsPreprint 2026
  5. Does SGD Seek Flatness or Sharpness? An Exactly Solvable ModelPreprint 2026
  6. Topological Invariance and Breakdown in Learning DynamicsNeurIPS 2026
  7. Machine Learning Decoding of Circuit-Level Noise for Bivariate Bicycle CodesQuantum 10, 2149 (2026)
  8. A universal compression theory for lottery ticket hypothesis and neural scaling lawsICLR 2026
All →

Neuroscience

  1. Connectome-based biophysical modeling of a figure-ground discrimination circuitPreprint 2026
  2. How Biological Synapses Self-Assemble Gradient LearningPNAS 2026
  3. Ubiquity of Emergent Hebbian Dynamics in Regularized LearningICML 2026
  4. Does Feedback Alignment Work at Biological Timescales?Preprint 2025
  5. Heterosynaptic Circuits Are Universal Gradient MachinesPreprint 2025
  6. Compositional Generalization via Forced Rendering of Disentangled LatentsICML 2025

AI

Neural Thermodynamics: Entropic Forces in Deep and Universal Representation Learning
  1. Learning the identity: a case study of how SGD selects among functional decompositionsPreprint 2026
  2. Width-Independent Compressibility of Deep Neural NetworksPreprint 2026
  3. Neural Quadratic Forms: A Unified Minimal Model for Sudden Learning and Scaling LawsPreprint 2026
  4. Thermodynamic Irreversibility of Training AlgorithmsPreprint 2026
  5. Does SGD Seek Flatness or Sharpness? An Exactly Solvable ModelPreprint 2026
  6. How Biological Synapses Self-Assemble Gradient LearningPNAS 2026
  7. Topological Invariance and Breakdown in Learning DynamicsNeurIPS 2026
  8. On the Invariance and Generality of Neural Scaling LawsNeurIPS 2026
All →

Talks I gave recently

  • Symmetry breaking in deep learning深層学習に於ける対称性の破れ
    Doctor thesis · 2023
  • Mean-field learning dynamics of deep neural networks
    Master thesis · 2020