We are currently working on a project where sources don't always have the same number of input neurons, and we are not in a position to calculate them easily by hand. We would find it quite useful to have a boolean toggle to the MLDataSet loaders like autodetect which, if turned on, basically assumes that (total num of columns in file) - (num of ideal columns) = the required number of input neurons. Could this be added to Encog? (Or is it in there already?)
We are currently working on a project where sources don't always have the same number of input neurons, and we are not in a position to calculate them easily by hand. We would find it quite useful to have a boolean toggle to the MLDataSet loaders like
autodetect
which, if turned on, basically assumes that (total num of columns in file) - (num of ideal columns) = the required number of input neurons. Could this be added to Encog? (Or is it in there already?)