神经网络进行mnist数据集识别总结

1、单个隐藏层神经网络构建：

程序:

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
print ("dajidali")
from tensorflow.examples.tutorials.mnist import input_data
mnist=input_data.read_data_sets('data/',one_hot=True)

//数据声明

numClasses=10
inputSize=784
numHiddenUnits=80
trainingIteratoins=50000
batchSize=100

变量声明
X=tf.placeholder(tf.float32,[None,inputSize])
y=tf.placeholder(tf.float32,[None,numClasses])

W1=tf.Variable(tf.truncated_normal([inputSize,numHiddenUnits],stddev=0.1))
B1=tf.Variable(tf.constant(0.1),[numHiddenUnits])
W2=tf.Variable(tf.truncated_normal([numHiddenUnits,numClasses],stddev=0.1))
B2=tf.Variable(tf.constant(0.1),[numClasses])

结构搭建

hiddenLayerOutput=tf.matmul(X,W1)+B1
hiddenLayerOutput=tf.nn.relu(hiddenLayerOutput)
finalOutput=tf.matmul(hiddenLayerOutput,W2)+B2
finalOutput=tf.nn.relu(finalOutput)

训练迭代

loss=tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y,logits=finalOutput))
opt=tf.train.GradientDescentOptimizer(learning_rate=.1).minimize(loss)

correct_prediction=tf.equal(tf.argmax(finalOutput,1),tf.argmax(y,1))
accuracy=tf.reduce_mean(tf.cast(correct_prediction,"float"))
init = tf.initialize_all_variables()
sess=tf.InteractiveSession()
sess.run(init)//初始化
for i in range(trainingIteratoins):
    batch=mnist.train.next_batch(batchSize)
    batchInput=batch[0]
    batchLabels=batch[1]
    _, trainingLoss=sess.run([opt,loss],feed_dict={X: batchInput,y: batchLabels})
    if i%1000==0:
        train_accuracy=accuracy.eval(session=sess,feed_dict={X:batchInput,y:batchLabels})
        print("step %d, training accuracy %g"%(i,train_accuracy))
batch = mnist.test.next_batch(batchSize)
testAccuracy = sess.run(accuracy, feed_dict={X:batch[0], y: batch[1]})
print("test accuracy %g"%(testAccuracy))

结果比对：1、numHiddenUnits=50，trainingIteratoins=10000时                

                                          

                2、numHiddenUnits=80，trainingIteratoins=50000

                                          

精度有所提升。

2、2个隐藏层数

                                         

          从第8000次开始训练精度达到很高的标准。

3、构建卷积神经网络在mnist数据集做训练

程序：

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
import random
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data', one_hot=True)

print ("dajidali")

tf.reset_default_graph()
sess=tf.InteractiveSession()

numHiddenUnits1=50
numHiddenUnits2=80
trainingIteratoins=5000
batchSize=50

X=tf.placeholder("float",shape=[None,28,28,1])
y_=tf.placeholder("float",shape=[None,10])

W_conv1=tf.Variable(tf.truncated_normal([5,5,1,32],stddev=0.1))
b_conv1=tf.Variable(tf.constant(0.1,shape=[32]))

h_conv1=tf.nn.conv2d(input=X, filter=W_conv1, strides=[1,1,1,1], padding='SAME')+b_conv1
h_conv1=tf.nn.relu(h_conv1)
h_pool1=tf.nn.max_pool(h_conv1,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

def conv2d(x, W):
    return tf.nn.conv2d(input=x, filter=W, strides=[1, 1, 1, 1], padding='SAME')
def max_pool_2x2(x):
    return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')

W_conv2=tf.Variable(tf.truncated_normal([5,5,32,64],stddev=0.1))
b_conv2=tf.Variable(tf.constant(0.1,shape=[64]))
h_conv2=tf.nn.relu(conv2d(h_pool1,W_conv2)+b_conv2)
h_pool2=max_pool_2x2(h_conv2)

W_fc1=tf.Variable(tf.truncated_normal([7*7*64,1024],stddev=0.1))
b_fc1=tf.Variable(tf.constant(0.1,shape=[1024]))
h_pool2_flat=tf.reshape(h_pool2,[-1,7*7*64])
h_fc1=tf.nn.relu(tf.matmul(h_pool2_flat,W_fc1)+b_fc1)

keep_prob=tf.placeholder("float")
h_fc1_drop=tf.nn.dropout(h_fc1,keep_prob)

W_fc2=tf.Variable(tf.truncated_normal([1024,10],stddev=0.1))
b_fc2=tf.Variable(tf.constant(0.1,shape=[10]))

y=tf.matmul(h_fc1_drop,W_fc2)+b_fc2

loss=tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_,logits=y))
opt=tf.train.AdamOptimizer().minimize(loss)

correct_prediction=tf.equal(tf.argmax(y,1),tf.argmax(y_,1))
accuracy=tf.reduce_mean(tf.cast(correct_prediction,"float"))

init =tf.global_variables_initializer()
sess.run(init)
for i in range(trainingIteratoins):
    batch=mnist.train.next_batch(batchSize)
    batchInput=batch[0].reshape([batchSize,28,28,1])
    batchLabels=batch[1]
    if i%100==0:
        train_accuracy=accuracy.eval(session=sess,feed_dict={X:batchInput,y_:batchLabels, keep_prob: 1.0})
        print("step %d, training accuracy %g"%(i,train_accuracy))
    opt.run(session=sess, feed_dict={X: batchInput,y_:batchLabels, keep_prob: 0.5})

batch = mnist.test.next_batch(batchSize)
batchInput=batch[0].reshape([batchSize,28,28,1])
batchLabels=batch[1]
testAccuracy = sess.run(accuracy, feed_dict={X:batchInput, y_: batchLabels,keep_prob: 1.0})
print("test accuracy %g"%(testAccuracy))

训练结果：

  至此mnist数据集上的小练习完成。