I design algorithms to understand the world.
My go-to metaphor is spatial.
I'm a partner at Probably Studio where we work primarily with computer vision techniques in diverse areas such as biomedical imaging and creative tools.
Previously, I worked on uncertainty modelling and the explainability of AI at Element AI. Still earlier I cofounded Whirlscape. We built Dango, using neural networks to embed emoji in 1000 dimensional semantic space. We built Minuum, searching for words in 10 dimensional keyboard space.
I love to explore the role of computation in creative work. If "We make our tools and our tools make us", so building computational models of reality causes us to experience it differently. For instance, I built a toy to simulate the refraction of light through curved surfaces. Now I notice these patterns everywhere.
Lately I've been using neural networks to create images, and I'm newly sensitive to the textures of the world.
In this spirit, I regularly give talks, teach workshops, organize art galleries at major computer vision conferences and organize the quasi-annual GenArtHackParty, where we teach people how to build generative art, and have a party to show it off. Many past winners had never programmed before, which is a point of pride.
This is incomplete and relatively infrequently updated, some work comes out via conferences, other gets posted with minimal documentation to my Twitter. Some work is also documented on my Google Scholar page.
- Multi-Scale Neural Texture Synthesis High-resolution synthesis! 2017
- Studies Visual Studies 2014
- Parallax Walls Passive re-lightable display technology (Disney Research) 2013
- Iso Isometric cubes 2013
- Old Toronto Maps from the Toronto Archives 2011
- Caustics Patterns with light 2010
- Ancient projects...
3D Coordinates by Oliviu Stoian from the Noun Project