I am a Flatiron Research Fellow in the NeuroAI and Geometric Data Analysis Lab led by SueYeon Chung at Flatiron Institute’s Center for Computational Neuroscience. I received my PhD in Physics in 2022 from Harvard University, where I worked on deep learning theory under the supervision of Cengiz Pehlevan. Before that, I earned my M.Sc. in Physics and B.Sc. in Electronics Engineering from Sabancı University.
Current Research
My research lies at the intersection of neuroscience, machine learning, and physics, with a focus on understanding how the brain processes and represents natural stimuli. I develop theoretical frameworks and computational models to study neural representations of sensory inputs. I apply tools from statistical mechanics, random matrix theory, and deep learning to uncover the principles underlying neural computation and perception.
I am particularly interested in:
- Neural representation of natural stimuli and their geometric structure
- Information processing in biological and artificial neural networks
- Development of robust tools for describing representational geometry
- Statistical physics of learning and neural computation
Selected Publications
Spectral Analysis of Representational Similarity with Limited Neurons
Hyunmo Kang*, Abdulkadir Canatar*, and SueYeon Chung
Advances in Neural Information Processing Systems, 2025 Link
Estimating Neural Representation Alignment from Sparsely Sampled Inputs and Features
Chanwoo Chun*, Abdulkadir Canatar*, SueYeon Chung, and Daniel Lee
8th Annual Conference on Cognitive Computational Neuroscience, 2025 Link
Statistical mechanics of support vector regression
Abdulkadir Canatar and SueYeon Chung
Physical Review E, 2025 Link
A Spectral Theory of Neural Prediction and Alignment
Abdulkadir Canatar*, Jenelle Feather*, Albert Wakhloo, and SueYeon Chung
Advances in Neural Information Processing Systems, 36, 2024 Link
Out-of-distribution generalization in kernel regression
Abdulkadir Canatar, Blake Bordelon, and Cengiz Pehlevan
Advances in Neural Information Processing Systems, 2021 Link
Spectral bias and task-model alignment explain generalization in kernel regression and infinitely wide neural networks
Abdulkadir Canatar, Blake Bordelon, and Cengiz Pehlevan
Nature Communications, 2021 Link
Contact
Office: 160 Fifth Ave, New York, NY 10010
Email: acanatar@flatironinstitute.org