def predict(self, x: np.array) -> np.array:
W1, W2 = self.params['W1'], self.params['W2']
b1, b2 = self.params['b1'], self.params['b2']
a1 = np.dot(x, W1) + b1
z1 = act.sigmoid(a1)
a2 = np.dot(z1, W2) + b2
y = act.softmax(a2)
return y
def predict(network, x):
W1, W2, W3 = network['W1'], network['W2'], network['W3']
b1, b2, b3 = network['b1'], network['b2'], network['b3']
a1 = np.dot(x, W1) + b1
z1 = act.sigmoid(a1)
a2 = np.dot(z1, W2) + b2
z2 = act.sigmoid(a2)
a3 = np.dot(z2, W3) + b3
y = act.softmax(a3)
return y
def gradient(self, x, t):
W1, W2 = self.params['W1'], self.params['W2']
b1, b2 = self.params['b1'], self.params['b2']
grads = {}
batch_num = x.shape[0]
# forward
a1 = np.dot(x, W1) + b1
z1 = act.sigmoid(a1)
a2 = np.dot(z1, W2) + b2
y = act.softmax(a2)
# backward
dy = (y - t) / batch_num
grads['W2'] = np.dot(z1.T, dy)
grads['b2'] = np.sum(dy, axis=0)
da1 = np.dot(dy, W2.T)
dz1 = act.sigmoid_grad(a1) * da1
grads['W1'] = np.dot(x.T, dz1)
grads['b1'] = np.sum(dz1, axis=0)
return grads
def forward(self, x: np.array, t: np.array) -> np.array:
self.t = t
self.y = act.softmax(x)
self.loss = loss.cross_entropy_error(self.y, self.t)
def loss(self, x: np.array, t: np.array) -> np.array:
z = self.predict(x)
y = act.softmax(z)
return loss.cross_entropy_error(y, t)