FAT-WiSARD

A Weighted Weightless Neural Network.

Depencies:

Python 3.X Numpy and Tensorflow

Docs:

Import:

from FATWiSARD import *

Create new Model:

model = FATWiSARD ( number_of_inputs, number_of_classes, number_of_rams, radius_initial_value = 1 )

Train:

model.train ( inputs, classes )

Opimize:

model.optimize ( inputs, classes, learning_rate, number_of_epochs = 1 )

Opitional Kwargs:
optimizer = "Adam" -> Uses Adam optimizer instead of Gradient Descent.
log = n -> Prints on the screen "Epoch: x" at a step of n.

OBS: classes can be one hot encoded or not.

Predict:

model.predict ( input )