def main():
Xtrain, Ytrain, Xvalid, Yvalid = getImageData()
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(Xtrain, Ytrain, Xvalid, Yvalid)
def main():
X, Y = getImageData()
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(X, Y)
def main():
X, Y = getImageData()
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(X, Y)
def main():
"""main"""
# Obtains 4D data for CNN
X, Y = getImageData()
model = CNN(convpool_layer_sizes=[(32, 5, 5), (64, 5, 5), (128, 5, 5)],
hidden_layer_sizes=[500, 300],
pool_sz=[(2, 2), (2, 2), (2, 2)])
model.fit(X, Y, show_fig=False)
def main(): X, Y = getImageData() model = CNN( convpool_layer_sizes=[(32, 3, 3), (64, 3, 3), (128, 3, 3), (256, 3, 3)], hidden_layer_sizes=[500, 300], ) model.fit(X, Y, display_cost=True)
def main():
X, Y = getImageData()
X = X.transpose(0, 2, 3, 1).astype(np.float32)
model = CNN(convpool_sz=[(20, 5, 5), (20, 5, 5)], hidden_sz=[500, 300])
t0 = datetime.now()
costs, acc = model.train(X, Y, epochs=500, batch_sz=100)
print('Time to train for 1000 epochs w/ batch sz 500: {}'.format(
datetime.now() - t0))
def main(argv): pw = argv[0] inputfile = argv[1] # print "password: "******"Input image: " + inputfile key = getKey(pw) # print "Key: " + key.hex data = getImageData(inputfile) bytes = decrypt(data, key) print bytes
def main():
X, Y = getImageData()
X = X.transpose((0, 2, 3, 1)) #0=N, 1-Color, 2-Width, 3-Height
#No of feature maps, feature width, feature height
# No of convpool layers=2
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(X, Y, show_fig=True)
def main():
Xtrain, Ytrain, Xvalid, Yvalid = getImageData()
# reshape X for tf: N x H x W x C
Xtrain = Xtrain.transpose((0, 2, 3, 1))
Xvalid = Xvalid.transpose((0, 2, 3, 1))
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(Xtrain, Ytrain, Xvalid, Yvalid)
def main():
X, Y = getImageData()
# reshape X for tf: N x w x h x c
X = X.transpose((0, 2, 3, 1))
print("X.shape:", X.shape)
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(X, Y)
def main():
X, Y = getImageData()
# reshape X for tf: N x w x h x c
X = X.transpose((0, 2, 3, 1))
print "X.shape:", X.shape
model = CNN(
convpool_layer_sizes=[(20, 5, 5), (20, 5, 5)],
hidden_layer_sizes=[500, 300],
)
model.fit(X, Y)
def main():
"""main"""
# Obtains 4D data for CNN
X, Y = getImageData()
# convert to tf [N,w,h,c]
X = X.transpose((0, 2, 3, 1))
print("X.shape:", X.shape)
print("Y.shape:", Y.shape)
model = CNN(
convpool_layer_sizes=[(32, 5, 5), (64, 5, 5), (128, 5, 5)],
strides=[[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]],
pool_sz=[[1, 2, 2, 1], [1, 2, 2, 1], [1, 2, 2, 1]],
pool_strides=[[1, 2, 2, 1], [1, 2, 2, 1], [1, 2, 2, 1]],
hidden_layer_sizes=[500, 300]
)
model.fit(X, Y, show_fig=False)
def main():
X, Y = getImageData()
# X, Y = shuffle(X, Y)
# Xtest, Ytest = X[-8000:,].astype(np.float32), Y[-8000:].astype(np.int32)
# X, Y = X[:-8000,].astype(np.float32), Y[:-8000].astype(np.int32)
# transpose the data in the tensorflow's order:
X = X.transpose((0, 2, 3, 1))
print('Training set shape:', X.shape)
# Xtest = Xtest.transpose((0, 2, 3, 1))
# print('Test set shape:', Xtest.shape)
model = CNN(
convpool_layer_sizes=[(32, 3, 3), (64, 3, 3), (128, 3, 3),
(256, 3, 3)],
hidden_layer_sizes=[500, 500],
)
model.fit(X, Y, display_cost=True)
# joblib.dump(model, 'mymodel.pkl')
# model = joblib.load('mymodel.pkl')
train_acc = model.score(X, Y)
print('\nTraining set acc: %.3f' % train_acc)
import numpy as np
import matplotlib.pyplot as plt
import cnn_tf as cnn
import tensorflow as tf
from util import getImageData
X, Y = getImageData()
X = X.transpose(0, 2, 3, 1).astype(np.float32)
learning_rates = [10e-3, 10e-4, 10e-5, 10e-6, 10e-7,
10e-8] # best learning rate is 10e-8
# learing_rate = 10e-8
# convLayer_sz = [[(10, 5, 5), (10, 5, 5)], [(20, 5, 5), (20, 5, 5)],
# [(30, 5, 5), (30, 5, 5)], [(40, 5, 5), (40, 5, 5)]]
convLayer_sz = [(10, 5, 5), (10, 5, 5)]
mlpLayer_sz = [500, 300]
costs = []
acc = []
file = open('Validation Data Set Scores.txt', 'w')
file.write('Validation results:\n')
for j, lr in enumerate(learning_rates):
# for i, cp in enumerate(convLayer_sz):
model = cnn.CNN(convpool_sz=convLayer_sz, hidden_sz=mlpLayer_sz)
c, a = model.train(X,
Y,
learning_rate=lr,
