Converts images into triangles, rectangles, and circles using Policy Gradients with Parameter-based Exploration (PGPE). It's an implementation of the "Fitting Concrete Target Image" portion of the ES-CLIP project.

A small experiment to create a "calculator language model" that can take in a string like "1 + 2" and output "3". This includes a system for generating arithmetic expressions of varying complexity to train the language model on.

A project using time series analysis to estimate and predict COVID-19 case rates. It included a dashboard that updated nightly from May 2020 to August 2022 and could pull in data from multiple sources.

An edge-detection library that shows how vector-order statistics can be used to turn the Canny edge detector from greyscale to RGB.

Originally started as a tool to setup a Traefik reverse proxy and a collection of Docker containers running on a single host. Really an experiment to learn about infrastructure automation.

An image clustering framework I created to try and find trends in images. It mainly uses Gaussian Mixture Models but I also began playing around with Variational Auto Encoders.

A simple app for generating photographic mosaics using the CIFAR-100 dataset for the image tiles. This came out of a photography course project, of all things.

A small experiment with PyTorch and the MNIST digits database. There's a Jupyter notebook in the repo that explains how it all works.

It's this website. Go to the GitHub repo to see how it's configured.