python程式碼

import tensorflow as tf 
import numpy as np 

from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('/tmp/data/', one_hot=True)

xtrain, ytrain = mnist.train.next_batch(2000)       # 用5000個樣本做容器物件
xtest, ytest = mnist.test.next_batch(200)           # 200個測試樣本
 


xi = tf.placeholder(tf.float32, [None, 784])        # 數值可變的變數都要宣告成Variable
yi = tf.placeholder(tf.float32,784)

distance = tf.reduce_sum(tf.abs(tf.add(xi, tf.negative(yi) )), reduction_indices=1 )
pred = tf.argmin(distance, 0)				# 最短的距離

init = tf.global_variables_initializer()

acc = 0.0
with tf.Session(
 
) as sess:
    sess.run(init)

    for i in range(len(xtest)):
        sample = xtest[i,:]
        pres_index = sess.run(pred, feed_dict={xi:xtrain, yi:sample })
        pred_label =  np.argmax(ytrain[pres_index])
        target = np.argmax(ytest[i])
        print('sample:{} | pred:{} | target:{}'.format
 
(i, pred_label, target))
        
        if pred_label == target:
            acc += 1/len(xtest)

    print('finished, final accuracy is:{}'.format(acc))

總結

1. tensorflow可以只進行前向計算，比如本例，相當於就是numpy