def new_neuralnet(train_set):
neural_net = NeuralNetwork()
# Adicionando a camada de input no camada 0
neural_net.camadas.append(
Layer(False, train_set.shape[1], train_set.shape[1]))
neural_net.functions.append(identidade)
neural_net.derivatives.append(identidade)
# Adicionando a camada escondida com 342 neurônios na camada 1
neural_net.camadas.append(Layer(True, train_set.shape[1], 342))
neural_net.functions.append(relu)
neural_net.derivatives.append(reluDerivative)
# Adicionando a camada escondida com 180 neurônios na camada 2
neural_net.camadas.append(Layer(True, 342, 180))
neural_net.functions.append(relu)
neural_net.derivatives.append(reluDerivative)
# Adicionando a camada de saída com 10 neurônios na camada 3
neural_net.camadas.append(Layer(True, 180, 10))
neural_net.functions.append(softmax)
neural_net.derivatives.append(softmax_derivative)
return neural_net
def new_neuralnet(train_set):
neural_softmax = NeuralNetwork()
# Adicionando a camada de input no indice 0
neural_softmax.camadas.append(Layer(False, train_set.shape[1], train_set.shape[1]))
neural_softmax.functions.append(identidade)
neural_softmax.derivatives.append(identidade)
# Adicionando a camada de saída com 10 neurônios no índice 1
neural_softmax.camadas.append(Layer(True, train_set.shape[1], 10))
neural_softmax.functions.append(softmax)
neural_softmax.derivatives.append(softmax_derivative)
return neural_softmax
def generate_layers_linear_regression():
input_neuron = Neuron('input_neuron', sigmoid)
output_neuron = Neuron('output_neuron', sigmoid)
independ_neuron = Neuron('independ_neuron', sigmoid, initial=1)
input_neurons = [input_neuron]
input_synapses = [
Synapse(input_neuron, output_neuron, 1.0),
]
input_layer = Layer('input_layer', input_neurons, input_synapses)
input_layer.set_linked_synapses()
layer_1_neurons = [independ_neuron]
layer_1_synapses = [
Synapse(independ_neuron, output_neuron, 1.0),
]
layer_1 = Layer('layer_1', layer_1_neurons, layer_1_synapses)
layer_1.set_linked_synapses()
output_neurons = [output_neuron]
output_synapses = [
Synapse(output_neuron, output_neuron, 1.0),
]
output_layer = Layer('output_layer', output_neurons, output_synapses)
output_layer.set_linked_synapses()
all_layers = [input_layer, layer_1, output_layer]
return all_layers
TODO: use GUI to generate input file
"""
import numpy as np
import matplotlib.pyplot as plt
from LayerClass import Layer
from ExperimentClass import Experiment
from Functions import SFCalc, MakeSequence, Post
_OutputFile = "../Projects/test1"
_iter = 10
# SAMPLE DEFINITION
# Sample = [vacuum , Top Layer, ..., Substrate]
Sample = [Layer(), Layer(), Layer(), Layer()]
# Sample[i].TakeValues[["Element"], nmultilayer,
# Sample.density,
# Sample.thick,
# Sample.roughness,
# Sample.MMC,
# Sample.phi,
# SAmple.gamma]
Sample[1].TakeValues([
"W", 10, [0.05, 0.0653, .12], [5., 10., 20.], [0, 0, 2], [0, 0, 1],
[0, 0, 90], [0, 0, 90]
])
Sample[2].TakeValues([
"Si", 10, [0.02, 0.05, .12], [0, 40, 60], [0, 0, 2], [0, 0, 1], [0, 0, 90],
def generate_layers_xor():
input_neuron_1 = Neuron('input_neuron_1', linear)
input_neuron_2 = Neuron('input_neuron_2', linear)
neuron_1_1 = Neuron('neuron_1_1', sigmoid)
neuron_1_2 = Neuron('neuron_1_2', sigmoid)
output_neuron = Neuron('output_neuron', sigmoid)
input_neurons = [input_neuron_1, input_neuron_2]
input_synapses = [
Synapse(input_neuron_1, neuron_1_1, 0.45),
Synapse(input_neuron_1, neuron_1_2, 0.78),
Synapse(input_neuron_2, neuron_1_1, -0.12),
Synapse(input_neuron_2, neuron_1_2, 0.13),
]
input_layer = Layer('input_layer', input_neurons, input_synapses)
input_layer.set_linked_synapses()
layer_1_neurons = [
neuron_1_1,
neuron_1_2,
]
layer_1_synapses = [
Synapse(neuron_1_1, output_neuron, 1.5),
Synapse(neuron_1_2, output_neuron, -2.3),
]
layer_1 = Layer('layer_1', layer_1_neurons, layer_1_synapses)
layer_1.set_linked_synapses()
output_neurons = [output_neuron]
output_synapses = [
Synapse(output_neuron, output_neuron, 1.0),
]
output_layer = Layer('output_layer', output_neurons, output_synapses)
output_layer.set_linked_synapses()
all_layers = [input_layer, layer_1, output_layer]
return all_layers