Tensorflow撸代码之3knn算法

具体knn算法概念参考knn代码python实现
上面是参考《机器学习实战》的代码，和knn的思想

# _*_ encoding=utf8 _*_

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

# 导入手写体识别的数据
mnist = input_data.read_data_sets("../data", one_hot=True)

# 训练集和测试集
X_train, Y_train = mnist.train.next_batch(5000) # 数据和labels
X_test, Y_test = mnist.test.next_batch(100)

# 定义输入
x_train = tf.placeholder(tf.float32, shape=(None,784))
x_test = tf.placeholder(tf.float32, shape=(784))

# L1距离也就是城市街区距离 |x1-x2|+|y1-y2|
distance = tf.reduce_sum(tf.abs(tf.add(x_train,tf.negative(x_test))),reduction_indices = 1)

# 返回最近的坐标，0纵轴 1横轴
pred = tf.arg_min(distance, 0)

accuracy = 0

# 初始化
init = tf.global_variables_initializer()

with tf.Session() as sess:

    sess.run(init)

    for i in range(len(X_test)):
        # 获取当前样本的最近邻索引，当前样本和每一个训练的样本找一个最近的l1距离，得到这个最小距离的下标
        nn_index = sess.run(pred, feed_dict={x_train:X_train, x_test: X_test[i, :]})

        # 由最邻近索引找到label,然后最邻近的label与真实标签比较 np.argmax找最大的下标
        # 由l1距离找到的最小值对应的坐标，通过该最坐标找到对应行label的最大值的下标，这个下标对应的就是数字的大小
        print("预测次数", i, "预测标签:", np.argmax(Y_train[nn_index]),"真实标签:", np.argmax(Y_test[i]))
        # 计算准确率
        if np.argmax(Y_train[nn_index]) == np.argmax(Y_test[i]):
            accuracy += 1

print("Accuracy:", float(accuracy)/len(X_test))