手动搭建神经网络实现mnist数据集的训练

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

载入数据集

mnist = input_data.read_data_sets(“mnist_data”, one_hot=True)

每个批次的大小

batch_size = 100

计算一共多少个批次

n_batch = mnist.train.num_examples // batch_size

定义两个placeholder 一张图片为784列个数据点

x = tf.compat.v1.placeholder(tf.float32, [None, 784])
y = tf.compat.v1.placeholder(tf.float32, [None, 10])

dropout 层的确定

keep_prob = tf.compat.v1.placeholder(tf.float32)

定义学习率

lr = tf.Variable(0.001, dtype=tf.float32, )

创建一个神经网络 产生正态分布组成的权重

W1 = tf.Variable(tf.random.truncated_normal([784, 500], stddev=0.1))
b1 = tf.Variable(tf.zeros([500]) + 0.1) # 偏置项

l1 = tf.nn.tanh(tf.matmul(x, W1) + b1)
l1_drop = tf.nn.dropout(l1, keep_prob)

W2 = tf.Variable(tf.random.truncated_normal([500, 300], stddev=0.1))
b2 = tf.Variable(tf.zeros([300]) + 0.1)
l2 = tf.nn.tanh(tf.matmul(l1_drop, W2) + b2)
l2_drop = tf.nn.dropout(l2, keep_prob)

W3 = tf.Variable(tf.random.truncated_normal([300, 10], stddev=0.1))
b3 = tf.Variable(tf.zeros(10) + 0.1)
prediction = tf.nn.softmax(tf.matmul(l2_drop, W3) + b3)

交叉熵损失

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y, logits=prediction))

采用Adam自适应学习率优化器

optimizer = tf.train.AdamOptimizer(lr)

最小化损失函数

train_step = optimizer.minimize(loss)

初始化

init_op = tf.global_variables_initializer()

求准确率的方法，结果放在一个布尔型列表中 argmax(y,1)表示按照行选取最大值

correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(prediction, 1))

求准确率

accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))

with tf.compat.v1.Session() as sess:
sess.run(init_op)
# 所有图片训练50次
for epoch in range(51):
# 每次epoch都主动改变学习率的大小
sess.run(tf.compat.v1.assign(lr, 0.001 * (0.95 ** epoch)))

    for batch in range(n_batch):
        batch_xs, batch_ys = mnist.train.next_batch(batch_size)
        sess.run(train_step, feed_dict={x: batch_xs, y: batch_ys, keep_prob: 0.8})

    learning_rate = sess.run(lr)
    acc = sess.run(accuracy, feed_dict={x: mnist.test.images, y: mnist.test.labels, keep_prob: 1.0})

    print("iter:{}, test accuracy:{}, learning_rate:{}".format(epoch, acc, learning_rate))