These MATLAB codes can be used to reproduce and extend the examples in the Numerical Methods for Deep Learning class.
These materials was first developed by Lars Ruthotto at Emory University and Eldad Haber at University of British Columbia in the Spring and Summer of 2018.
The MATLAB files here aim at implementing of neural networks and numerical methods for their training in the simplest possible way. This is critical here, in order to involve students and researchers from a variety of backgrounds. It is more important to us than speed in this case. In fact, many functions here can be optimized for runtime but often at the cost of less readability.
Development of this material is in part supported by the National Science Foundation under Grant Numbers 1522599 and 1751636. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.