def main2():
infrastructure = np.array([2, 3, 2])
initial_values = np.array([1, 0])
network = neural.Network()
nodeCount = infrastructure.sum()
for i in range(0, infrastructure.__len__()):
for layerCount in range(0, infrastructure[i]):
node = neural.Node()
if i == 0:
node.value = initial_values[i]
node.bias = randint(-10, 10)
network.add_node("layer" + str(i + 1) + "node" + str(layerCount),
node)
network.show_network()
# -*- coding: utf-8 -*-
"""
Created on Tue Mar 6 14:14:54 2018
@author: jia335
Some default brains to start with
"""
import neuralnetwork as NN
'''
The basic brain I used at the start. Not very clever.
'''
simpleBrain = NN.Network()
simpleBrain.createNetwork([3, 2], [2, 2])
'''
Moves towards nearest food (hopefully)
'''
naiveBrain = NN.Network()
naiveBrain.addInputs(3)
naiveBrain.addOutputs(2)
naiveBrain.addSynapse(naiveBrain.inputs[1], naiveBrain.outputs[1])
naiveBrain.addSynapse(naiveBrain.inputs[0], naiveBrain.outputs[0])
for syn in naiveBrain.synapses:
syn.weight = 1
for out in naiveBrain.outputs:
out.type = 'Sum'
brains = {'Simple Brain': simpleBrain, 'Naive Brain': naiveBrain}
import mnist_loader
import numpy as np
import matplotlib.pyplot as plt
import neuralnetwork as network
def plot(pixels):
pixels = np.array(pixels, dtype='float32')
# Reshape the array into 28 x 28 array (2-dimensional array)
pixels = pixels.reshape((28, 28))
# Plot
plt.imshow(pixels, cmap='gray')
plt.show()
training_data, validation_data, test_data = mnist_loader.load_data_wrapper()
arr, label = test_data[-1]
net = network.Network([784, 100, 10])
net.default_weight_initializer()
net.stochastic_gradient_descent(training_data,
30,
10,
0.5,
evaluation_data=test_data,
monitor_evaluation_accuracy=True)
print net.predict(arr)
ball = Ball(WHITE) ball.rect.x = 240 ball.rect.y = 150 all_sprites_list.add(ball) ball_list.add(ball) # Loop until the user clicks the close button. done = False # Used to manage how fast the screen updates clock = pygame.time.Clock() pop=[neuralnetwork.Network([5,3]) for i in range(20)] brain=neuralnetwork.createandtrain(pop,100) score_reached = False # Setting the score display score1 = 0 score1_font = pygame.font.Font(None, 50) score1_surf = score1_font.render(str(score1) , 1, (255, 255, 255)) score1_pos = [30, 10] score2 = 0 score2_font = pygame.font.Font(None, 50) score2_surf = score2_font.render(str(score2), 1, (255, 255, 255)) score2_pos = [360, 10]
import eigen_face_loader trainingData, testData = eigen_face_loader.load_data() import neuralnetwork net = neuralnetwork.Network([len(trainingData), 200, 15]) net.SGD(trainingData, 50, 5, .09, 15, test_data = testData)
import neuralnetwork
import pickle
#Creating, Training and Saving
network = neuralnetwork.Network([2, 3, 1], [0.2, 0.1], 0.5)
andGateTrainingSet = {
(0, 0): (0, ),
(0, 1): (0, ),
(1, 0): (0, ),
(1, 1): (1, )
}
network.debug(True) #periodically shows the average error of the training set
network.training(andGateTrainingSet)
pickle.dump(network, open("filename.p", "wb"))
#Loading and Using
loadedNetwork = pickle.load(open("filename.p", "rb"))
loadedNetwork.forwardPass((0, 1))
result = loadedNetwork.getOutput()
print(
result
) #outputs [0.001156001503972413] which is close to the intended output (0)
''' import mnist_loader trainingData, validation_data, testData = mnist_loader.load_data_wrapper() import copyCode net = copyCode.Network([len(trainingData), 100, 10]) net.SGD(trainingData, 100, 30, .1, test_data = testData) ''' import mnist_loader trainingData, test_data = mnist_loader.load_data_wrapper() import neuralnetwork net = neuralnetwork.Network([784, 30, 10]) net.SGD(trainingData, 10, 10, 3.0, 10, test_data=test_data)
import neuralnetwork as net import mnist_loader #loading MNIST handwritten number image data training_data, validation_data, test_data = mnist_loader.load_data_wrapper() training_data = list(training_data) #Pre-training visual example of data mnist_loader.show_data() #Creating network with 784 input sigmoid neurons, 30 hidden, 10 output net = net.Network([784, 30, 10]) net.SGD(training_data, 30, 10, 3.0, test_data=test_data)