Correlation flow governs learning at criticality
Connecting the flow of correlations through deep networks to their learning dynamics at criticality.
A new initialisation to Control Gradients in Sinusoidal Neural network
A way to initialise sinusoidal networks that controls gradients and improves how they learn functions and images.
Nonlinear effects in fluctuation short pulse reflectometry
Simulating how microwaves scatter in turbulent plasma, with machine learning to improve turbulence measurements.
