I'm thinking about the geometry and topology of neural networks.
Tooling for neural networks.
Orchestrate massive neural network architecture searches.
Interactive convnet features visualization
Declarative hyperparameter experiments for neural networks.
Realtime monitoring for neural network training
Mostly some review notes
Deep Learning According to a Differential Geometer
Braids and Mapping Class Groups
Introduction to Calibrated Manifolds
The Moduli Space measure in bosonic String Theory
Interactive, machine-learning powered product recommendation graphs.
Versatile window tiling for X11 with powerful keyboard controls.
Graphics and physics engine for arbitrary dimensions.