Using computational neuroscience, AI, and data science to understand how we see.
Associate Research Scientist, Columbia University, Visual Inference Lab.
I research new, general methods of data visualization which I use to reveal how neural networks and human brains solve visual tasks.
A computational neuroscientist trained in neural network and mathematical modeling, machine learning, and electrophysiology, I approach problems with rigor and an eye for experimentation.
My toolkit includes:
Python, PyTorch, MATLAB, deep learning, machine learning, computer vision, high performance (cluster) computing, electrophysiology.
current* and past work
hypersphere2sphere*
A novel and general framework for dimensionality-reducing visualization, hypersphere2sphere takes high-dimensional distributions of labeled data as objects and shows their relationships.
disentangling neural networks with dynamics*
Convolutional neural networks can achieve amazing performance on many computer vision tasks, but are brittle in the face of carefully selected, imperceptibly small changes to images ("adversarial attacks"). Networks could be made more robust with inductive biases that come from the statistics of simple dynamic worlds.
cortical sheet network*
A novel, brain-inspuired neural network architecture based on biological feature map organization.
unifying theory of motion, fit to data
Previous theories of visual motion computation by neurons in area MT could not capture the wide temporal frequency tuning they commonly exhibit. A series of targeted experiments fit with theoretical models revealed the importance of tuned normalization.
hyperplaids
Time for running experiments is limited. By prioritizing regions of spatiotemporal frequency space, the types of motion that excite and suppress MT neurons could be measured in unprecedented detail.
motion subunits
How motion tuning changes in different spatial locations of MT receptive fields was poorly understood. A novel filtered-noise stimulus revealed this spatial organization.
extracellular recording
Autobiography of an electrode.