def predict(self, x):
w1, w2 = self.params['w1'], self.params['w2']
b1, b2 = self.params['b1'], self.params['b2']
z1 = np.dot(x, w1) + b1
a1 = sigmoid(z1)
z2 = np.dot(a1, w2) + b2
y = softmax(z2)
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 = sigmoid(a1)
a2 = np.dot(z1, w2) + b2
y = 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 = sigmoid_grad(a1) * da1
grads['w1'] = np.dot(x.T, dz1)
grads['b1'] = np.sum(dz1, axis=0)
return grads
import numpy as np
from common.funcs import softmax, softmax2
a = [[1, 2, 3, 4, 5],
[3, 5, 6, 8, 9],
[0, 7, 2, 4, 6]]
a = np.array(a)
print(softmax(a) == softmax2(a))
def forward(self, x, t):
self.t = t
self.y = softmax(x)
self.loss = cross_entropy_error(self.y, self.t)
return self.loss
def loss(self, x, t):
z = self.predict(x)
y = softmax(z)
loss = cross_entropy_error(y, t)
return loss