Dendrite-Neural-Networks 0.0.5

Dendrite-Neural-Networks is an implementation of processing units that performed classification using closed decision boundaries with only one neuron.

Dendrite-Neural-Networks

Dendrite-Neural-Networks is a Python library for dealing with closed boundary classification units, like Dendrite Morphological Neuron (DMN), Dendrite Ellipsoidal Neuron (DEN) and Dendrite Spherical Neuron (DSN).

Installation

Use the package manager pip to install Dendrite-Neural-Networks.

pip install Dendrite-Neural-Networks

Usage

from DEN import DENlayer
from DMN import DMNlayer
from DSN import DSNlayer

from PreTrain.HpC import HSpC
from PreTrain.HpC import HEpC
from PreTrain.HpC import HBpC

from PreTrain.kmeans import bkmeans
from PreTrain.kmeans import ekmeans
from PreTrain.kmeans import skmeans

# returns 'array' with the propose initial parameters
#x: Input pattern
#y: Labels
dendrites  = HBpC.HBpC(x,y,0.0001)
dendrites  = HEpC.HEpC(x,y)
dendrites  = HSpC.HSpC(x,y,0.0001)
dendrites  = bkmeans.bkmeans(x,y,[3,3,3],0.01)
dendrites  = ekmeans.ekmeans(x,y,[3])
dendrites  = skmeans.skmeans(x,y,[2],0.01)


# It's an implementation of a modified Keras layer
DMNlayer(2,dendrites, activation = "sigmoid", input_shape = (np.shape(x)[1],)))
DENlayer(2,dendrites, activation = "sigmoid", input_shape = (np.shape(x)[1],)))
DSNlayer(2,dendrites, activation = "sigmoid", input_shape = (np.shape(x)[1],)))

# It's an implementation of a modified Keras layer. For random initialization of parameters
DMNlayer(2, activation = "sigmoid", input_shape = (np.shape(x)[1],)))
DENlayer(2, activation = "sigmoid", input_shape = (np.shape(x)[1],)))
DSNlayer(2, activation = "sigmoid", input_shape = (np.shape(x)[1],)))

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

GNU GPLv3