David G. Clark

David G. Clark

dgc2138@cumc.columbia.edu

Graduate Student in Neurobiology & Behavior @ Columbia University

Hello!

I am a neuroscience PhD candidate at Columbia University, primarily advised by Larry Abbott. I work in the Center for Theoretical Neuroscience.

Some characteristic features of neural circuits are 1) there are a lot of neurons, 2) neurons are nonlinear, 3) neurons engage in complex recurrent dynamics, and 4) connections between neurons change on a variety of timescales. In addition to making neural circuits hard to understand, these features underlie computation and learning. A major challenge is to understand how this works. My research approaches this challenge using tools from statistical physics and machine learning.

My publications are listed below, or see Google Scholar.

Education

  • PhD in Neurobiology & Behavior

    Columbia University, 2019–Present

  • B.A. in Physics, Computer Science

    UC Berkeley, 2017

Publications

(2024). Simplified derivations for high-dimensional convex learning problems. arXiv.

arXiv

(2024). Connectivity structure and dynamics of nonlinear recurrent neural networks. arXiv.

arXiv

(2024). Theory of coupled neuronal-synaptic dynamics. Physical Review X.

PDF Journal Physics Magazine Viewpoint arXiv

(2024). Structure of activity in multiregion recurrent neural networks. arXiv.

arXiv

(2023). Dimension of activity in random neural networks. Physical Review Letters.

PDF Journal arXiv

(2021). Olfactory landmarks and path integration converge to form a cognitive spatial map. Neuron.

PDF Journal Code Video

(2021). Credit assignment through broadcasting a global error vector. NeurIPS 2021.

arXiv Code

(2019). Unsupervised discovery of temporal structure in noisy data with dynamical components analysis. NeurIPS 2019.

arXiv Code

(2017). Neuromorphic Kalman filter implementation in IBM's TrueNorth. Journal of Physics: Conference Series.

PDF Journal