Tensorflow2.0 label与one-hot独热编码向量之间的相互转换

Tensorflow2.0 label与one-hot独热编码向量之间的相互转换_清风思月的博客-CSDN博客https://blog.csdn.net/zhaoguanghe/article/details/103442630

目录

1.label 转 one-hot

2.one-hot 转 label

---

1.label 转 one-hot

 
import tensorflow as tf
 
label = tf.stack(5)
one_hot_label = tf.one_hot(label, num_classes)
 
print("label: ", label.numpy())
print("one_hot_label: ", one_hot_label.numpy())
 
'''
output:
label: 5
one_hot_label: [0., 0., 0., 0., 0., 1., 0., 0., 0., 0.]
'''
 
 

2.tensor 类型one-hot 转 label

label = tf.stack(5)
one_hot_label = tf.one_hot(label, 10)
 
new_label = tf.argmax(one_hot_label)
print(new_label.numpy())
 
# output: 5

3.ndarray类型one-hot 转 label

labels.argmax(axis=1)