course

:: exploring compressed sensing fMRI time series

4 minute read Published:

we are once again solving systems of equations
An exploration of compressed sensing fMRI time series with 3 different algorithms. Typically, compressed sensing reconstructs a single volume of MRI but fMRI are composed of many volumes; sensing along the time domain could reduce the number of volumes required. Of the 3 algorithms, BSBL-BO performed the best with the error curve elbowing around 30% subsampling.

:: verification of a VAE & SegNet using NNV

6 minute read Published:

''Many things that seem threatening in the dark become welcoming when we shine light on them.'' -- Uncle Iroh
Neural network automated verification of a VAE and SegNet using NNV. Although neural networks are promising, they are easily confused, particularly if the input domain is perturbed. In this project, I demonstrate the robustness of MNIST-trained VAE and SegNet against varying brightness attacks.

:: visualizing temporal graph networks

5 minute read Published:

me, 6 months ago: data viz is a dedicated research field??, me, now: thank gawd for data viz researchers...
Visualizing the resulting link prediction graph from a temporal graph network on a Wikipedia dataset using Observable and d3.js.

:: face following and vSLAM for a Tello quadcopter

7 minute read Published:

Tello can do hard things.
Implementation of face detection / following and vSLAM on a Ryze Tello using its MATLAB toolkit.

:: dimensionality reduction on neural data

5 minute read Published:

PCA vs. autoencoder: the ultimate dimensionality reduction showdown
I fell in love with dimensionality reduction when I was learning statistical ML. Since I also study neuroscience, I wanted to practice the art at the intersection of my interests. I compared the 3D projections of a 53-dimensional neurophysiology dataset produced by PCA and a shallow autoencoder.