def forward(self, x, t):
self.t = t
self.y = 1 / (1 + np.exp(-x))
self.loss = cross_entropy_error(np.c_[1 - self.y, self.y], self.t)
return self.loss
def estimate_E_out(f, w, n_rounds=1000):
'''
Estima o erro fora da amostra de h[w](x), gerando uma nova amostra
e calculando o erro de entropia cruzada sobre ela.
'''
dataset = generate_dataset(f, n_rounds)
return myround(cross_entropy_error(w, dataset))
def loss(self, x, t):
z = self.predict(x)
#y = softmax(z)
y = functions.relu(z)
loss = functions.cross_entropy_error(y, t)
return loss
def forward(self, x: np.ndarray, t: np.ndarray) -> float:
self.t = t
self.y = softmax(x)
if self.t.size == self.y.size:
self.t = self.t.argmax(axis=1)
loss = cross_entropy_error(self.y, self.t)
return loss
def forward(self, x, t):
self.t = t
self.y = softmax(x)
if self.t.size == self.y.size:
self.t = self.t.argmax(axis=1)
loss = cross_entropy_error(self.y , self.t)
return loss
def forward(self, x, t):
self.t = t
self.y = softmax(x)
# 教師ラベルがone-hotベクトルの場合、正解のインデックスに変換
if self.t.size == self.y.size:
self.t = self.t.argmax(axis=1)
loss = cross_entropy_error(self.y, self.t)
return loss
def forward(self, x, t):
self.t = t
self.y = softmax(x)
# 정답 레이블이 원핫 벡터일 경우 정답의 인덱스로 변환
if self.t.size == self.y.size:
self.t = self.t.argmax(axis=1)
loss = cross_entropy_error(self.y, self.t)
return loss
def loss(self, x, t): """ x : array-like input t : array-like true label """ z = self.predict(x) y = softmax(z) loss = cross_entropy_error(y, t) return loss
def loss(self, x, t):
"""Loss function using cross entropy error
Args:
x (numpy.ndarray): image data which mean input to NN
t (numpy.ndarray): labels
Returns:
float: result of cross entropy error
"""
y = self.predict(x)
return cross_entropy_error(y, t)
def forward(self, x, t):
"""forward propagation
Args:
x (numpy.ndarray): input
t (numpy.ndarray): train data
Returns:
float: cross entropy error
"""
self.t = t
self.y = softmax(x)
self.loss = cross_entropy_error(self.y, self.t)
return self.loss
def forward(self, x, t):
"""順伝播
SoftmaxWithLossレイヤの順伝播の結果を返す
Args:
x (ndarray): 入力
t (ndarray): 教師ラベル
"""
self.t = t
self.y = softmax(x)
# 教師ラベルがone-hotベクトルの場合、正解のインデックスに変換
if self.t.size == self.y.size:
self.t = self.t.argmax(axis=1)
loss = cross_entropy_error(self.y, self.t)
return loss
def loss(self, x, t):
y = self.predict(x)
return f.cross_entropy_error(y, t)
def forward(self, x, true):
self.true = true
self.pred = f.softmax(x)
self.loss = f.cross_entropy_error(self.pred, self.true)
return self.loss
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
def loss(self,x,t): #lossの算出 by cross_entropy_error
y = self.predict(x)
return cross_entropy_error(y,t)
import numpy as np
"""データ"""
x = np.array([[0.1, 0.8]])
w1 = np.array([[10, 7], [0.8, 6]])
b1 = np.array([[1, 1]])
w2 = np.array([[0.4, 30], [0.8, 0.2]])
b2 = np.array([[1, 1]])
t = np.array([[1, 0]])
learning_rate = 0.02
for i in range(1):
a1 = np.dot(x, w1) + b1
z1 = relu(a1)
a2 = np.dot(z1, w2) + b2
y = softmax(a2)
loss = cross_entropy_error(y, t)
"""共通の偏微分"""
dEdY = -(t / y) # [dEdy1, dEdy2]
S = np.sum(np.exp(a2))
dYdS = np.exp(a2) / np.square(S) # [dY1dS, dY2dS]
print("dw2_11: ",
(-(t[0][0] / y[0][0]) *
(np.exp(a2[0][1]) / np.square(S))) * np.exp(a2[0][0]) * z1[0][0])
print("dw2_12: ",
(-(t[0][0] / y[0][0]) *
(np.exp(a2[0][1]) / np.square(S))) * np.exp(a2[0][1]) * z1[0][0])
print("dw2_21: ",
(-(t[0][0] / y[0][0]) *
(np.exp(a2[0][1]) / np.square(S))) * np.exp(a2[0][0]) * z1[0][1])
print("dw2_22: ",
def evaluate_E_in(w, dataset):
'''
Calcula o erro dentro da amostra (rotulada) dataset, de uma hipótese parametrizada por w.
'''
return myround(cross_entropy_error(w, dataset))
def loss(self, x, t): #xは入力データ，tは正解ラベル
z = self.predict(x)
y = softmax(z)
loss = cross_entropy_error(y, t)
return loss
def forward(self, x: np.ndarray, t: np.ndarray) -> np.ndarray:
self.t = t
self.y = 1 / (1 + np.exp(-x))
self.loss = cross_entropy_error(np.c_[1 - self.y, self.y], self.t)
return self.loss
def loss(self, t):
self.t = t
self.error = cross_entropy_error(self.y, self.t)
return self.error
import numpy as np
import matplotlib.pylab as plt
import sys,os
sys.path.append("../lib")
sys.path.append(os.pardir)
import functions
from dataset.mnist import load_mnist
t = [0,0,1,0,0,0,0,0,0,0]
y_good = [0.1, 0.05, 0.6, 0.0, 0.05, 0.1, 0.0, 0.1, 0.0, 0.0]
print(functions.mean_squared_error(np.array(y_good), np.array(t)))
y_bad = [0.1, 0.05, 0.2, 0.0, 0.05, 0.1, 0.0, 0.1, 0.8, 0.0]
print(functions.mean_squared_error(np.array(y_bad), np.array(t)))
print(functions.cross_entropy_error(np.array(y_good), np.array(t)))
print(functions.cross_entropy_error(np.array(y_bad), np.array(t)))
def loss(self, y, t):
# loss를 계산
error = cross_entropy_error(y, t)
return error
