def main():
X, Y = get_facialexpression(balance_ones=True)
M = 2000
ann = AnnTheano(M)
ann.fit(X, Y)
print "score:", ann.score(X, Y)
def main():
X, Y = get_facialexpression(balance_ones=True)
M = 200
ann = ANN(M)
ann.fit(X, Y, show_figure=True)
print ann.score(X, Y)
def main():
X, Y = get_facialexpression(balance_ones=True)
X, Y = shuffle(X, Y)
K = len(np.unique(Y))
#Split into train and test
Ntrain = int(0.8 * len(Y))
Xtrain, Ytrain = X[:Ntrain, :], Y[:Ntrain]
Xtest, Ytest = X[:Ntrain, :], Y[:Ntrain]
M = 100
# create the neural network
model = MLPClassifier(hidden_layer_sizes=(Ntrain, M),
activation='logistic',
learning_rate='constant',
learning_rate_init=1e-7,
verbose=True)
#train ANN
model.fit(Xtrain, Ytrain)
# print the train and test accuracy
train_accuracy = model.score(Xtrain, Ytrain)
test_accuracy = model.score(Xtest, Ytest)
print "train accuracy:", train_accuracy, "test accuracy:", test_accuracy
def main():
X, T = get_facialexpression(balance_ones=True)
# X, T = np.shuffle(X,T)
label_map = [
'Anger', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral'
]
# klass =3 error_rate=0.0
# klass =4 error_rate=0.0
# klass =5 error_rate=0.0
# klass =0
klass = 4
N, D = X.shape
X = np.concatenate(
(np.ones((N, 1)), X),
axis=1,
)
T = T.astype(np.int32)
X = X.astype(np.float32)
#Fix for forecasting on one image
T = class1detect(T, detect=klass)
D += 1
# params
lr = 5e-7
max_iteration = 150
W = np.random.randn(D) / np.sqrt(D)
cost = []
error = []
for i in xrange(max_iteration):
Y = forward(W, X)
cost.append(cross_entropy(T, Y))
error.append(error_rate(T, Y))
W += lr * X.T.dot(T - Y)
if i % 5 == 0:
print "i=%d\tcost=%.3f\terror=%.3f" % (i, cost[-1], error[-1])
if i % 5 == 0:
print "i=%d\tcost=%.3f\terror=%.3f" % (i, cost[-1], error[-1])
print "Final weight:", W
print T
print np.round(Y)
plt.title('logistic regression ' + label_map[klass])
plt.xlabel('iterations')
plt.ylabel('cross entropy')
plt.plot(cost)
plt.show()
plt.title('logistic regression ' + label_map[klass])
plt.xlabel('iterations')
plt.ylabel('error rate')
plt.plot(error)
plt.show()
def main():
print 'Loading ...'
X, Y = get_facialexpression(balance_ones=True)
detect = 1
print 'detecting class', detect
Y = class1detect(Y, detect)
model = LogisticModel()
model.fit(X, Y, epochs=5000, show_figure=True)
model.score(X, Y)
def main():
print 'Loading ...'
X, Y = get_facialexpression(balance_ones=True)
label_map = [
'Anger', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral'
]
print Y.shape
print Y
# detect=1
# print 'detecting class', detect
# Y = class1detect(Y, detect)
model = LogisticModelK(label=label_map)
model.fit(X, Y, epochs=200, show_figure=True)
model.score(X, Y)
def main():
label_map = [
'Anger', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral'
]
X, Y = get_facialexpression(balance_ones=False)
while True:
for i in xrange(7):
x, y = X[Y == i], Y[Y == i]
N = len(x)
j = np.random.choice(N)
plt.imshow(x[j].reshape(48, 48),
cmap='gray') # images have been flattened
plt.title(label_map[y[j]])
plt.show()
prompt = raw_input('Quit? Enter Y:\n')
if prompt == 'Y':
break
def main(): X, Y = get_facialexpression(balance_ones=True) M = 5000 ann = AnnTensorflow1(M) ann.fit(X, Y, learning_rate=10e-7, reg=10e-6, show_fig=True)
def main(): X, Y = get_facialexpression(balance_ones=True) ann = AnnTheano2([2000, 100, 500]) ann.fit(X, Y) print "score:", ann.score(X,Y)
def main():
X, Y = get_facialexpression(balance_ones=True)
ann = AnnTensorflow2([2000, 1000, 500])
print type(ann)
ann.fit(X, Y, show_figure=True)
def main(): X, Y = get_facialexpression(balance_ones=True) model = LogisticModelSoftmax() model.fit(X, Y, show_figure=True) print model.score(X, Y)