Tensorflow One-Hot 函数样例(Py3)

admin

March 16, 2018, 9:07 p.m.

read: 801

Py3代码：

# one-hot 函数的样例
import tensorflow as tf

label = tf.placeholder(tf.int32,[None])
# 直接把 输入的序列进行One-Hot的结果
one_hot = tf.one_hot(label, 3, 1, 0)
# 进行转置
one_hot_new = tf.transpose(one_hot, perm=[1,0])
one_hot_new = tf.cast(one_hot_new, tf.float32)
# one_hot_new[2] = one_hot_new[2] * 1.5

# 按照每两维的大小进行拆分
one_hot_new_1 = tf.dynamic_partition(one_hot_new, [0, 0, 1], 2)[0]
one_hot_new_2 = tf.dynamic_partition(one_hot_new, [0, 0, 1], 2)[1]
# 拼接以上两维得到原来的结果
one_hot_new = tf.concat([one_hot_new_1, one_hot_new_2], axis=0)


if __name__ == '__main__':
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        one_hot_out, one_hot_new_out, one_hot_new_1_out, one_hot_new_2_out = sess.run([one_hot, one_hot_new, one_hot_new_1, one_hot_new_2], feed_dict={label: [0, 1, 1, 2, 2, 0, 0, 1, 2, 2, 0, 2]})
        print("原始的One-hot结果:")
        print(one_hot_out, end='\n\n')
        print("以上的结果.T:")
        print(one_hot_new_out, end='\n\n')
        print("拆分（1,2）维:")
        print(one_hot_new_1_out, end='\n\n')
        print("拆分 (3)维:")
        print(one_hot_new_2_out, end='\n\n')

控制台输出:

原始的One-hot结果:
[[1 0 0]
[0 1 0]
[0 1 0]
[0 0 1]
[0 0 1]
[1 0 0]
[1 0 0]
[0 1 0]
[0 0 1]
[0 0 1]
[1 0 0]
[0 0 1]]

以上的结果.T:
[[ 1. 0. 0. 0. 0. 1. 1. 0. 0. 0. 1. 0.]
[ 0. 1. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]
[ 0. 0. 0. 1. 1. 0. 0. 0. 1. 1. 0. 1.]]

拆分（1,2）维:
[[ 1. 0. 0. 0. 0. 1. 1. 0. 0. 0. 1. 0.]
[ 0. 1. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]

拆分 (3)维:
[[ 0. 0. 0. 1. 1. 0. 0. 0. 1. 1. 0. 1.]]

附录（一个奇葩的方法）：

Py3:

import tensorflow as tf

labels = tf.placeholder(tf.float32)
labels_0 = 0.5 - labels
labels_0 = tf.nn.relu(labels_0) * 2

labels_1_t = labels - 1.5
labels_1_t = tf.nn.relu(labels_1_t) * 3
labels_1 = labels - 0.5
labels_1 = tf.nn.relu(labels_1)
labels_1 = labels_1 - labels_1_t
labels_1 = labels_1 * 2
labels_1 = tf.nn.relu(labels_1)

labels_2 = labels - 1.5
labels_2 = tf.nn.relu(labels_2)
labels_2 = labels_2 * 2

with tf.Session() as sess:
    print(sess.run([labels_0, labels_1, labels_2], feed_dict={labels: [0, 1, 2, 2, 2, 0, 1, 2, 0]}))

控制台：

[array([ 1., 0., 0., 0., 0., 1., 0., 0., 1.], dtype=float32), array([ 0., 1., 0., 0., 0., 0., 1., 0., 0.], dtype=float32), array([ 0., 0., 1., 1., 1., 0., 0., 1., 0.], dtype=float32)]