def AddOutLayer(self, cNeurons):
layer = ntl.NetLayer()
cIn = len(self.layers[len(self.layers) - 1].neurons) + 1
for i in range(cNeurons):
layer.addNeuron(cIn)
print("Agregando capa de salida...")
self.layers.append(layer)
def neuRandom(self, cIn, cLayers, cNeurons):
for i in range(cLayers):
layer = ntl.NetLayer()
self.layers.append(layer)
for j in range(0, cNeurons):
self.layers[i].addNeuron(cIn + 1)
if i > 0:
cIn = len(self.layers[i - 1].neurons)
def load(file):
#instanciamos la red
neuralNet = nn.NeuralNetwork()
with open(file) as oFile:
#leemos el archivo en formato JSON
jsonF = json.load(oFile)
#Iteramos por capas y neuronas
for layer in jsonF['capas']:
nLayer = ntl.NetLayer()
for neuron in layer['neuronas']:
w = neuron['pesos']
#Añadimos la neurona como arreglos [baias, arregloPesos, g(n), delta)
nLayer.neurons.append([1, w, 0, 0])
#Añadimos la capa con las neuronas a la red
neuralNet.layers = np.append(neuralNet.layers, nLayer)
return neuralNet
def load(file):
neuralNet = nn.NeuralNetwork()
with open(file) as oFile:
jsonF = json.load(oFile)
for layer in jsonF['capas']:
nLayer = ntl.NetLayer()
for neuron in layer['neuronas']:
print(neuron['pesos'])
w = neuron['pesos']
nLayer.neurons.append([1, w, 0, 0])
neuralNet.layers = np.append(neuralNet.layers, nLayer)
for i in range(0, 2):
for j in range(0, 2):
vector = [i, j]
neuralNet.evaluar(vector)
neuralNet.printG(vector)
def evaluar(self, vector):
ban = 0
before = ntl.NetLayer()
for layer in self.layers:
for neuron in layer.neurons:
x = 0.0
if ban != 0:
x = neuron[0] * neuron[1][0]
for i in range(len(before.neurons)):
x += neuron[1][i + 1] * before.neurons[i][2]
else:
x = neuron[0] * neuron[1][0]
for i in range(len(vector)):
x += neuron[1][i + 1] * vector[i]
neuron[2] = layer.sigmoide(x)
ban = 1
before = layer
def AddHiddenLayer(self, cNeurons, cIn):
layer = ntl.NetLayer()
for i in range(cNeurons):
layer.addNeuron(cIn + 1)
print("Agregando Capa oculta...")
self.layers.append(layer)