mcnets 1.2.1

A package that demonstrates deep neural nets using Monte Carlo-type parameter training.

Monte-Carlo-Neural-Nets

A package meant to demonstrate how well deep neural nets could be with random weight assignment / training (hence monte-carlo). Similar to popular machine learning packages in python, the created nets can consist of many layers, all custom in size, with different activation functions in between each layer to achieve the desire results (note the curve fitting example on the github page below).

By either having some data set to fit a model to, or by having some 'score' factor, the nets can be trained in a large variety of situations. For example, curve fitting, playing Snake, playing Chess, etc., have all successfully been done so far.

Some examples of the net's operation and training can be found on the GitHub page, where issues are also tracked: https://github.com/SciCapt/Monte-Carlo-Neural-Nets