浅谈keras中loss与val_loss的关系

时间:2022-09-01 17:23:51

loss函数如何接受输入值

keras封装的比较厉害,官网给的例子写的云里雾里,

在*找到了答案

You can wrap the loss function as a inner function and pass your input tensor to it (as commonly done when passing additional arguments to the loss function).

?
1
2
3
4
def custom_loss_wrapper(input_tensor):
 def custom_loss(y_true, y_pred):
  return K.binary_crossentropy(y_true, y_pred) + K.mean(input_tensor)
 return custom_loss
?
1
2
3
4
5
input_tensor = Input(shape=(10,))
hidden = Dense(100, activation='relu')(input_tensor)
out = Dense(1, activation='sigmoid')(hidden)
model = Model(input_tensor, out)
model.compile(loss=custom_loss_wrapper(input_tensor), optimizer='adam')

You can verify that input_tensor and the loss value will change as different X is passed to the model.

?
1
2
3
4
5
6
X = np.random.rand(1000, 10)
y = np.random.randint(2, size=1000)
model.test_on_batch(X, y) # => 1.1974642
 
X *= 1000
model.test_on_batch(X, y) # => 511.15466

fit_generator

fit_generator ultimately calls train_on_batch which allows for x to be a dictionary.

Also, it could be a list, in which casex is expected to map 1:1 to the inputs defined in Model(input=[in1, …], …)

?
1
2
3
4
### generator
yield [inputX_1,inputX_2],y
### model
model = Model(inputs=[inputX_1,inputX_2],outputs=...)

补充知识:学习keras时对loss函数不同的选择,则model.fit里的outputs可以是one_hot向量,也可以是整形标签

我就废话不多说了,大家还是直接看代码吧~

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
from __future__ import absolute_import, division, print_function, unicode_literals
import tensorflow as tf
from tensorflow import keras
import numpy as np
import matplotlib.pyplot as plt
 
print(tf.__version__)
fashion_mnist = keras.datasets.fashion_mnist
 
(train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()
class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
    'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']
# plt.figure()
# plt.imshow(train_images[0])
# plt.colorbar()
# plt.grid(False)
# plt.show()
 
train_images = train_images / 255.0
test_images = test_images / 255.0
 
# plt.figure(figsize=(10,10))
# for i in range(25):
#  plt.subplot(5,5,i+1)
#  plt.xticks([])
#  plt.yticks([])
#  plt.grid(False)
#  plt.imshow(train_images[i], cmap=plt.cm.binary)
#  plt.xlabel(class_names[train_labels[i]])
# plt.show()
 
model = keras.Sequential([
 keras.layers.Flatten(input_shape=(28, 28)),
 keras.layers.Dense(128, activation='relu'),
 keras.layers.Dense(10, activation='softmax')
])
 
model.compile(optimizer='adam',
    loss='categorical_crossentropy',
    #loss = 'sparse_categorical_crossentropy' 则之后的label不需要变成one_hot向量,直接使用整形标签即可
    metrics=['accuracy'])
one_hot_train_labels = keras.utils.to_categorical(train_labels, num_classes=10)
 
model.fit(train_images, one_hot_train_labels, epochs=10)
 
one_hot_test_labels = keras.utils.to_categorical(test_labels, num_classes=10)
test_loss, test_acc = model.evaluate(test_images, one_hot_test_labels)
 
print('\nTest accuracy:', test_acc)
 
# predictions = model.predict(test_images)
# predictions[0]
# np.argmax(predictions[0])
# test_labels[0]

loss若为loss=‘categorical_crossentropy', 则fit中的第二个输出必须是一个one_hot类型,

而若loss为loss = ‘sparse_categorical_crossentropy' 则之后的label不需要变成one_hot向量,直接使用整形标签即可

以上这篇浅谈keras中loss与val_loss的关系就是小编分享给大家的全部内容了,希望能给大家一个参考,也希望大家多多支持服务器之家。

原文链接:https://blog.csdn.net/u013608336/article/details/82559469