Abdulkadir Canatar
Education
Harvard University | Cambridge, MA
Ph.D. in Physics | 2016–2022
- Thesis: Statistical mechanics of generalization in kernel regression and wide neural networks
Sabancı University | Istanbul, Turkey
M.Sc. in Physics | 2015–2016
- Thesis: Fabrication and characterization of suspended graphene devices
Sabancı University | Istanbul, Turkey
B.Sc. in Electronics Engineering | 2010–2015
Research & Professional Experience
Flatiron Institute – Simons Foundation | New York, NY
Flatiron Research Fellow | 2022 – Present
- Supervisor: Prof. SueYeon Chung
Harvard University | Cambridge, MA
Teaching & Research Assistant | 2017 – 2022
- Supervisor: Prof. Cengiz Pehlevan
Sabancı University | Istanbul, Turkey
Undergraduate & Masters Researcher | 2012 – 2016
- Supervisor: Prof. İsmet İnönü Kaya
Publications
Equal Contribution indicated by asterisk (*)
Sample-Size-Invariant Measure of Dimensionality
Chanwoo Chun*, Abdulkadir Canatar*, SueYeon Chung and Daniel D. Lee — 2025 (In Submission)Spectral Analysis of Representational Similarity with Limited Neurons
Hyunmo Kang*, Abdulkadir Canatar* and SueYeon Chung — NeurIPS, 2025Estimating Neural Representation Alignment from Limited Inputs and Features
Chanwoo Chun*, Abdulkadir Canatar*, SueYeon Chung and Daniel D. Lee — CCN, 2025Statistical Mechanics of Support Vector Regression
Abdulkadir Canatar and SueYeon Chung — Physical Review E, 2025A Spectral Theory of Neural Prediction and Alignment
Abdulkadir Canatar, Jenelle Feather, Albert Wakhloo, and SueYeon Chung — NeurIPS, 2023 (Spotlight)Bandwidth Enables Generalization in Quantum Kernel Models
Abdulkadir Canatar, Evan Peters, Cengiz Pehlevan, Stefan M. Wild and Ruslan Shaydulin — TMLR, 2023A Kernel Analysis of Feature Learning in Deep Neural Networks
Abdulkadir Canatar and Cengiz Pehlevan — 58th Annual Allerton Conference, 2022Asymptotics of representation learning in finite Bayesian neural networks
Jacob Zavatone-Veth, Abdulkadir Canatar, Benjamin S. Ruben and Cengiz Pehlevan — NeurIPS, 2021Out-of-Distribution Generalization in Kernel Regression
Abdulkadir Canatar, Blake Bordelon and Cengiz Pehlevan — NeurIPS, 2021Spectral Bias and Task-Model Alignment Explain Generalization in Kernel Regression and Infinitely Wide Neural Networks
Abdulkadir Canatar, Blake Bordelon and Cengiz Pehlevan — Nature Communications, 2021Strong localization in suspended monolayer graphene by intervalley scattering
Cenk Yanik, Vahid Sazgari, Abdulkadir Canatar and İsmet İ. Kaya — Physical Review B, 2021Spectrum Dependent Learning Curves in Kernel Regression and Wide Neural Networks
Blake Bordelon, Abdulkadir Canatar and Cengiz Pehlevan — ICML, 2020
Conference Talks/Posters
Spectral Analysis of Representational Similarity with Limited Neurons
COSYNE 2025 (Poster)A Spectral Theory of Neural Prediction and Alignment
COSYNE 2024 (Poster)Out-of-Distribution Generalization in Kernel Regression
INFORMS 2023 (Invited Talk)Statistical Mechanics of Generalization in Kernel Regression and Wide Neural Networks
APS March Meeting 2022 (Talk)Statistical Mechanics of Generalization in Kernel Regression
DeepMath 2020 (Talk)A theory of generalization in kernel regression and wide neural networks
Neuromatch 2020 (Talk)
Teaching
Harvard University | Teaching Assistant
APMATH 50 - Introduction to Applied Mathematics
(Spring 2020–2021, Spring 2021–2022)APMATH 226 - Neural Computation
(Fall 2020–2021, Fall 2021–2022)PS 2 - Mechanics, Elasticity, Fluids, and Diffusion
(Fall 2019–2020)PS 12a - Mechanics from an Analytic and Numerical Perspective
(Spring 2018–2020)PHYS 232 - Advanced Electromagnetism
(Fall 2017–2018)
Technical Skills
- Programming: Python (PyTorch, JAX, NumPy, SciPy, Pandas), Mathematica, SLURM
- Research: Statistical physics, computational neuroscience, machine learning theory
Awards & Scholarships
Derek Bok Center (Harvard) – Certificate of Distinction in Teaching Awards (Multiple Terms)
Harvard University – Purcell Fellowship (2016–2017)
Sabancı University – Dean’s High Honor List & Honors Scholarship (2010–2016)