问题遇到的现象和发生背景
用代码块功能插入代码，请勿粘贴截图
import numpy as np
import matplotlib
matplotlib.use('TkAgg')
# import matplotlib.pyplot as plt
# x_train = pd.read_csv('D:\pycharm文件\轴承\轴承数据集/trdata.txt', delimiter=' ', header=None)
# data1 = pd.read_csv('D:\pycharm文件\轴承\轴承数据集/trlabel.txt', delimiter=' ')
# y_train = data1.iloc[0:4, :].values.T
# x_test = pd.read_csv('D:\pycharm文件\轴承\轴承数据集/txdata.txt', delimiter=' ', header=None)
# data2 = pd.read_csv('D:\pycharm文件\轴承\轴承数据集/txlabel.txt', delimiter=' ')
# y_test = data2.iloc[0:4, :].values.T
x_train = np.loadtxt('D:\pycharm文件\轴承\轴承数据集/trdata.txt', delimiter=' ')
y_ytrain = np.loadtxt('D:\pycharm文件\轴承\轴承数据集/trlabel.txt', delimiter=' ')
y_train = y_ytrain[0:4, 0:]
x_test = np.loadtxt('D:\pycharm文件\轴承\轴承数据集/txdata.txt', delimiter=' ')
y_ytest = np.loadtxt('D:\pycharm文件\轴承\轴承数据集/txlabel.txt', delimiter=' ')
y_test = y_ytest[0:4, 0:]
# def sigmoid(x):
# return 1/(1+np.exp(-x))
def sigmoid(x):
if x >= 0: #对sigmoid函数的优化，避免了出现极大的数据溢出
return 1.0/(1+np.exp(-x))
else:
return np.exp(x)/(1+np.exp(x))
def dsigmoid(y):
return y * (1-y)
n, m = np.shape(x_train)
n_x = 400
n_h1 = 600
n_h2 = 200
n_h3 = 50
n_y = 4
w1 = np.random.randn(n_h1, n_x)*0.1 # 600*400
b1 = np.zeros((n_h1, 1)) # 600*1
w2 = np.random.randn(n_h2, n_h1)*0.1 # 200*600
b2 = np.zeros((n_h2, 1)) # 200*1
w3 = np.random.randn(n_h3, n_h2)*0.1 # 50*200
b3 = np.zeros((n_h3, 1)) # 50*1
w4 = np.random.randn(n_y, n_h3)*0.1 # 4*50
b4 = np.zeros((n_y, 1)) # 4*1
def forward(x_train, w1, b1, w2, b2, w3, b3, w4, b4):
z1 = np.dot(w1, x_train) + b1 # 600*400 * 400*8000 =600 * 8000
a1 = sigmoid(z1) # 600 * 8000
z2 = np.dot(w2, a1) + b2 # 200*600 * 600*8000 = 200*8000
a2 = sigmoid(z2) # 200*8000
z3 = np.dot(w3, a2) + b3 # 50*200 * 200*8000 = 50*8000
a3 = sigmoid(z3) # 50*8000
z4 = np.dot(w4, a3) + b4 # 4*50 * 50*8000 = 4*8000
a4 = z4 # 4*8000
return z1, z2, z3, z4, a1, a2, a3, a4
def costfuction(a4, y_train):
n, m = np.shape(x_train)
error = np.sum(0.5 * (a4 - y_train.T) ** 2)/m
return error
def backforward(y_train, a4, a3, a2, a1, w4, w3, w2):
dz4 = a4 - y_train.T # 4*8000
dw4 = np.dot(dz4, a3.T) # 4*8000 * 8000*50 = 4*50
db4 = np.sum(dz4, axis=1, keepdims=True)/m # 4*8000
dz3 = np.dot(w4.T, dz4) * sigmoid(a3) # 50*4 * 4*8000 = 50*8000
dw3 = np.dot(dz3, a2.T) # 50*8000 * 8000*200 = 50*200
db3 = np.sum(dz3, axis=1, keepdims=True)/m # 50*8000
dz2 = np.dot(w3.T, dz3) * sigmoid(a2) # 200*50 * 50*8000 = 200*8000
dw2 = np.dot(dz2, a1.T) # 200*8000 * 8000*600 = 200*600
db2 = np.sum(dz2, axis=1, keepdims=True)/m # 200*8000
dz1 = np.dot(w2.T, dz2) * sigmoid(a1) # 600*200 * 200*8000 = 600*8000
dw1 = np.dot(dz1, x_train.T) # 600*8000 * 8000*400 = 600*400
db1 = np.sum(dz1, axis=1, keepdims=True)/m # 600*8000
return dz4, dw4, db4, dz3, dw3, db3, dz2, dw2, db2, dz1, dw1, db1
alpha = 0.01
number = 5000
for i in range(1, number+1):
z1, z2, z3, z4, a1, a2, a3, a4 = forward(x_train, w1, b1, w2, b2, w3, b3, w4, b4)
error = np.sum(0.5 * (a4 - y_train.T) ** 2) / m
dz4, dw4, db4, dz3, dw3, db3, dz2, dw2, db2, dz1, dw1, db1 = backforward(y_train, a4, a3, a2, a1, w4, w3, w2)
w1 = w1 - alpha * dw1
w2 = w2 - alpha * dw2
w3 = w3 - alpha * dw3
w4 = w4 - alpha * dw4
b1 = b1 - alpha * db1
b2 = b2 - alpha * db2
b3 = b3 - alpha * db3
b4 = b4 - alpha * db4
if (i % 1000 == 0):
print(i)
# plt.plot(i, error, 'ro')
# plt.show()
z1_test, z2_test, z3_test, z4_test, a1_test, a2_test, a3_test, a4_test = forward(x_test, w1, b1, w2, b2, w3, b3, w4, b4)
print(a4_test)
n_test, m_test = np.shape(x_test)
error_test = np.sum(0.5 * (a4_test - y_test.T) ** 2) / m_test
print(error_test)
c = 0
b = np.rint(a4_test.T) ## 对矩阵取整
n_2, m_2 = np.shape(a4_test.T)
for i in range(0, m_2):
c1 = b[i, :] ## 对两个矩阵相对应的位置作比较
c2 = y_test[i, :]
if (c1 == c2).any(): ## 不加all有3个tf，if函数识别不了三个，必须加all才能明白这个一整对矩阵是否相同
c = c + 1
acc = c / m_test * 100
print('正确了', c, "个")
print('准确率：%.2f%%' % acc)
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
我已经加了all()，为什么还是报这个错误
