Here are some of the projects I’ve been working on.
Using deterministic fluids estimation to encode parameters of a radiation beam
I used these inputs to train a neural net to produce Monte Carlo-derived dose-distributions in artificial geometries. This network was then used in a patient sample with moderate success.
A faster way to irradiate
Calculating radiation dose as quickly as Collapsed Cone Convolution, but as accurately as Monte Carlo.
There’s always a bigger fish
Koi is a project that uses a continuous genetic algorithm to optimise the neural movement-based decision-making of fish in a pond.
I heard that you like dimensions, so I put a dimension inside your dimension. Two doesn’t cut it for you? Have a third. More like three doesn’t cut it. Have a forth. Hyper Forest is a 2D render of a 3D slice of a 4D world. Welcome to the descent.
I. Have. Decision. Fatigue.
Lifebot was only ever supposed to be a personal project. I tried to turn depression into a technical problem so that I could solve it. Life has a high degree of equifinality. It didn’t matter when I did it so I just didn’t. Now the bot tells me when to do it and now I’m… fine I guess. Full Disclosure : I do not recommend this bot. Just go outside.
*Northern Running Figures
Style can only exist in the context of other styles. I made a bot that compares things and produced a relative special-temporal chronology for indigenous rock art on the basis of indigenous art from the Northern Arnhem Lands.
Let the robot take the wheel
Phase 1: Previous authors have achieved AI piloted flat drone flight - we will replicate this. Phase 2: We will give the AI more directions.
Robots CAN make beautiful masterpieces
Pose estimation via videography as an input for a neural net, output to Project Magenta to make some banging tunes.
Living vicariously through algorithms because I don’t have time to play video games anymore.