Batch and instance whitening
This short post will cover graphical intuition and PyTorch code for two different kinds of whitening: batch and instance.
about me
I'm a computational & theoretical neuroscience PhD student at NYU advised by Eero Simoncelli and David Heeger. My dissertation focuses on theories of adaptive efficient coding in recurrent neural networks, and developing statistical tools for analyzing neural representational geometry.
In the summer of 2022, I was a PhD intern on the Open Codecs team at Google where I did a research project on machine learning for video compression with adaptive nonlinear autoencoders.
I was born and raised in Yellowknife, Northwest Territories, Canada 🥶🍁. My BSc is in Physiology and Physics from McGill University, where I began working on what would eventually become my MSc at the University of Western Ontario (our lab migrated), modeling neural correlates of visuo-spatial attention in prefrontal cortex, advised by Julio Martinez-Trujillo.
hear me pronounce my Vietnamese family name, Dương: Saying "Yuh-ng" will get you close (the D sounds like a Y in English). Most people pronounce it like "Dew-ong" or "Dwong", which I'm fine with.
recent notes
Orthogonal iteration
A tutorial for a classic numerical linear algebra algorithm to find eigenvectors and values.
Elliptical slice sampling
A Python implementation of the elegant algorithm introduced by Iain Murray et al. (2010).
Interpolating covariances with their geometric mean
What’s the best way to quantify and visualize distance between two positive definite matrices? Julia code included.
Stochastic gradient Markov chain Monte Carlo
PyTorch implementation and explanation of SGD MCMC sampling w/ Langevin or Hamiltonian dynamics.