-
Getting good signal flow in wide neural networks
To train wide networks effectively, there's only one effective degree of freedom in choosing how hyperparameters scale.
-
How to take derivatives of matrix expressions
Matrix differentiation doesn't have to be painful.
-
Quick n dirty derivation of Larmor radiation and gravitational waves
It's not so hard to derive order-of-magnitude expressions for radiation formulas
-
Statistical physics of signal propagation in deep neural networks
Using statistical physics to understand what makes a good activation function
-
Automatically detecting and tracking structures in protostellar outflows
My senior thesis won an award