Demonstrate a simple linear neural network with bandicoot
This is another proof-of-concept to make sure things are headed in the right direction.
-
make sure ensmallen works with bandicoot (#9 (closed)) -
implement a multilayer linear feedforward neural network with a simple nonlinearity like a sigmoid for bandicoot outside of mlpack's framework, directly with ensmallen -
demonstrate speedup vs. arma::mat
for that simple neural network -
implement the same multilayer linear feedforward neural network with mlpack's framework (which likely needs to be modified somewhat)