daviddoji/vanilla-machine-learning
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

learning for both nn working

Browse Source
This commit is contained in:
Ritchie
2017-07-01 12:01:07 +02:00
parent ea28e2ee29
commit 43a8098031
2 changed files with 37 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
functions.py
View File

@@ -238,8 +238,9 @@ class Network:
:param dw: (array) Partial derivatives :param dw: (array) Partial derivatives
:param delta: (array) Delta error. :param delta: (array) Delta error.
""" """
self.w[index] -= self.learning_rate * np.mean(dw, 1)
self.b[index] -= self.learning_rate * np.mean(np.mean(delta, 1), 0) self.w[index] -= self.learning_rate * dw
self.b[index] -= self.learning_rate * np.mean(delta, 0)
def fit(self, x, y_true, loss, epochs, batch_size, learning_rate=2e-2): def fit(self, x, y_true, loss, epochs, batch_size, learning_rate=2e-2):
""" """
@@ -265,7 +266,8 @@ class Network:
self.back_prop(z, a, y_[k:l]) self.back_prop(z, a, y_[k:l])
if (i + 1) % 10 == 0: if (i + 1) % 10 == 0:
print("Loss:", self.loss.loss(y_true, z[self.n_layers])) _, a = self.feed_forward(x)
print("Loss:", self.loss.loss(y_true, a[self.n_layers]))
def predict(self, x): def predict(self, x):
_, a = self.feed_forward(x) _, a = self.feed_forward(x)
@@ -273,27 +275,37 @@ class Network:
if __name__ == "__main__": if __name__ == "__main__":
from sklearn import datasets from sklearn import datasets
#import sklearn.metrics import sklearn.metrics
np.random.seed(1) np.random.seed(1)
# Load data # Load data
data = datasets.load_iris() data = datasets.load_iris()
x = data["data"] x = data["data"]
x = (x - x.mean()) / x.std() x = (x - x.mean()) / x.std()
y = np.expand_dims(data["target"], 1) y = data["target"]
#y = np.expand_dims(data["target"], 1)
# one hot encoding # one hot encoding
y = np.eye(3)[y] y = np.eye(3)[y]
nn = Network((4, 8, 2, 3), (Relu, Relu, Sigmoid)) nn = Network((4, 8, 3), (Relu, Relu, Sigmoid))
#nn.fit(x[:2], y[:2], MSE, 1, batch_size=2) #nn.fit(x[:2], y[:2], MSE, 1, batch_size=2)
nn.fit(x, y, MSE, 1000, 16) nn.fit(x, y, MSE, 1000, 16)
# data = datasets.load_digits()
#
# x = data["data"]
# y = data["target"]
# y = np.eye(10)[y]
#
# nn = Network((64, 32, 10), (Relu, Sigmoid))
# nn.fit(x, y, MSE, 100, 2)
#
y_ = nn.predict(x) y_ = nn.predict(x)
a = np.argmax(y_, 1) a = np.argmax(y_, 1)
for i in range(a.size): # for i in range(a.size):
print(a[i], y[i]) # print(a[i], y[i], "\t", np.round(y_[i], 3))
# y_true = [] # y_true = []
# y_pred = [] # y_pred = []

32
simple_mlp.py
View File

@@ -140,10 +140,10 @@ class NeuralNetwork:
dc_dw1 = np.dot(self.x.T, delta2) dc_dw1 = np.dot(self.x.T, delta2)
# update weights and biases # update weights and biases
self.w[2] -= self.learning_rate * np.mean(dc_dw2, 1) self.w[2] -= self.learning_rate * dc_dw2
self.b[2] -= self.learning_rate * np.mean(np.mean(delta3, 1), 0) self.b[2] -= self.learning_rate * np.mean(delta3, 0)
self.w[1] -= self.learning_rate * np.mean(dc_dw1, 1) self.w[1] -= self.learning_rate * dc_dw1
self.b[1] -= self.learning_rate * np.mean(np.mean(delta2, 1), 0) self.b[1] -= self.learning_rate * np.mean(delta2, 0)
def stats(self): def stats(self):
""" """
@@ -188,24 +188,24 @@ if __name__ == "__main__":
data = datasets.load_iris() data = datasets.load_iris()
x = data["data"] x = data["data"]
x = (x - x.mean()) / x.std() x = (x - x.mean()) / x.std()
y = np.expand_dims(data["target"], 1) y = data["target"]
# one hot encoding # one hot encoding
y = np.eye(3)[y] y = np.eye(3)[y]
nn = NeuralNetwork(4, 8, 3, 2e-2) nn = NeuralNetwork(4, 4, 3, 1e-2)
#nn.fit(x[:2], y[:2], 2, 1) #nn.fit(x[:2], y[:2], 2, 1)
nn.fit(x, y, 16, 1000) nn.fit(x, y, 8, 1000)
_, y_ = feed_forward(x, nn.w, nn.b) _, y_ = feed_forward(x, nn.w, nn.b)
print(y_) print(y_[3])
# # result # # result
# _, y_ = feed_forward(x, nn.w, nn.b) # _, y_ = feed_forward(x, nn.w, nn.b)
# y_true = [] y_true = []
# y_pred = [] y_pred = []
# for i in range(len(y)): for i in range(len(y)):
# y_pred.append(np.argmax(y_[3][i])) y_pred.append(np.argmax(y_[3][i]))
# y_true.append(np.argmax(y[i])) y_true.append(np.argmax(y[i]))
#
# print(sklearn.metrics.classification_report(y_true, y_pred)) print(sklearn.metrics.classification_report(y_true, y_pred))
#