Colloquium: What Kinds of Functions Do Neural Networks Learn? Low-Norm vs. Flat Solutions
Rahul Parhi, UC San Diego
This talk investigates the fundamental differences between low-norm and flat solutions of shallow ReLU networks training problems,...
Colloquium: Nodal statistics of graphs and matrices
John Urschel, MIT
The study of discrete nodal statistics, that is, data regarding the zeros of Laplacian eigenvectors, provides insight into structural...
