ffnet is meant to be an easy and fast feed-forward neural network training solution for python.
Here are some key features of "ffnet":
· Any network connectivity without cycles is allowed.
· Training can be performed with use of several optimization schemes including genetic alorithm based optimization.
· There is access to exact partial derivatives of network outputs vs. its inputs.
· Automatic normalization of data.
ffnet uses some external open source software:
· networkx for handling network architecture as the graph data structure,
· numpy/scipy for training purposes,
· pikaia, genetic algorihm based oprimizer in fortran (included).
· matplotlib for plotting (examples only).