def btnRun_Clicked(self):
if self.image == '':
msg_box_img = QMessageBox()
msg_box.setWindowTitle("Warning")
msg_box_img.setIcon(QMessageBox.Question)
msg_box_img.setText(
'The test image had not been loaded.\n Would you like to use the default image?'
)
msg_box_img.setStandardButtons(QMessageBox.Yes | QMessageBox.No)
img_retval = msg_box_img.exec_()
if img_retval == QMessageBox.Yes:
self.image = DEFAULT_IMAGE
self.pimage = QPixmap(self.image)
self.limage.setPixmap(self.pimage)
if self.model == '':
msg_box_model = QMessageBox()
msg_box.setWindowTitle("Warning")
msg_box_model.setText(
'The model had not been loaded.\n Would you like to use the default model?'
)
msg_box_model.setIcon(QMessageBox.Question)
msg_box_model.setStandardButtons(QMessageBox.Yes | QMessageBox.No)
model_retval = msg_box_model.exec_()
if model_retval == QMessageBox.Yes:
self.model = DEFAULT_MODEL
modelpx = QPixmap("images/model.png")
modelpx = modelpx.scaled(256, 256)
self.lmodel.setPixmap(modelpx)
msg_box = QMessageBox()
msg_box.setWindowTitle("Finish")
if self.image != '' and self.model != '':
cnn = neuralNetwork(self.image, self.model)
self.y_preds = cnn.predict()
real_lm = revertPredict(self.y_preds)
self.print_prediction(real_lm)
self.glLabel.setPixmap(self.pimage)
msg_box.setText("Finish process !")
else:
msg_box = QMessageBox()
msg_box.setText("The image and the model do not empty !")
msg_box.exec_()
NOS_EPISODES = 250
SAVE_PATH = "./models/"
if not os.path.exists(SAVE_PATH):
os.mkdir(SAVE_PATH)
RENDER_GAME = False
# Setting up Gym
env = gym.make(ENV_NAME)
logger = benchmark()
# Finding the Network Shape
action_space = env.action_space.n
observation_space = env.observation_space.shape
# Initializing neural network
nn_solver = neuralNetwork(action_space, observation_space)
# Running Q Learning
for episode in range(NOS_EPISODES):
state = env.reset()
state = np.reshape(state, (1, observation_space[0]))
time_step = 0
terminate = False
while True:
# Render Game
if RENDER_GAME:
env.render()
# Increment Time Step
time_step = time_step + 1
# Take Action
action = nn_solver.act(state, env)
def show_image(data):
values = data.split(',')
image_array = np.asfarray(values[1:]).reshape(28, 28)
plt.imshow(image_array, cmap='Greys', interpolation='None')
plt.show()
# number of input nodes is a number of pixels
input_nodes = 784
hidden_nodes = 100
# each output node corresponds to particular number
output_nodes = 10
learning_rate = 0.3
# number of training epochs
epochs = 2
n = neuralNetwork(input_nodes, hidden_nodes, output_nodes, learning_rate)
# load train data
train_data_list = load_train_data(train_data_location)
# start training
print('Start training')
for i in range(epochs):
for sample in train_data_list:
#split sample
values = sample.split(',')
# scaling of inputs to the range (0.01, 1)
inputs = np.asfarray(values[1:])/255.0 * 0.99 +0.01
# choose 0.01 target value for outputs that doesn't correspond to label
targets = np.zeros(output_nodes) + 0.01
# and 0.99 otherwise
import neural_network as nk import numpy import matplotlib.pyplot as pl import scipy.special import scipy.ndimage.interpolation import json import time import progressbar # 测试神经网络 if __name__ == "__main__": input_nodes = nk.input_nodes hidden_nodes = nk.hidden_nodes hidden_nodes_2 = nk.hidden_nodes_2 #hidden_nodes_3 = nk.hidden_nodes_3 output_nodes = nk.output_nodes learningrate = nk.learningrate Network = nk.neuralNetwork(input_nodes,hidden_nodes,hidden_nodes_2,output_nodes,learningrate) nk.test(Network,"mnist_test.csv") # hidden_nodes = 200 lr = 0.01 performance = 97.34
from neural_network import neuralNetwork
import numpy as np
num_input_nodes = 784
num_hidden_nodes = 200 # random value -> we can choose the best value by heuristic
num_output_nodes = 10 # label is between 0 ~ 9
learning_rate = 0.1 # random value -> we can choose the best value by heuristic
neural_network = neuralNetwork(num_input_nodes, num_hidden_nodes,
num_output_nodes, learning_rate)
def training():
training_data_file = open("../mnist_dataset/mnist_train_100.csv", 'r')
training_data_list = training_data_file.readlines()
training_data_file.close()
epochs = 5 # random value -> we can choose the best value by heuristic
print("Start training!")
for e in range(epochs):
for record in training_data_list:
all_values = record.split(',')
inputs = (np.asfarray(all_values[1:]) / 255.0 * 0.99
) + 0.01 # adjust input value to between 0.01 and 0.99
targets = np.zeros(num_output_nodes) + 0.01 # To avoid 0
targets[int(all_values[0])] = 0.99 # answer target
