Molecular dynamics (MD) simulations allow atomistic insights into chemic...
Recent results suggest that reinitializing a subset of the parameters of...
Existing work on understanding deep learning often employs measures that...
We study deep neural networks (DNNs) trained on natural image data with
...
Assume we have potential "causes" z∈ Z, which produce "events" w with
kn...
We consider the core reinforcement-learning problem of on-policy value
f...
Temporal-Difference learning (TD) [Sutton, 1988] with function approxima...
Deep neural networks are often trained in the over-parametrized regime (...