人工智能从入门到精通(15)-卷积网络在数字识别的应用
来源:互联网 发布:淘宝返场是什么意思 编辑:程序博客网 时间:2024/05/17 01:35
经典卷积网络模型
LeNet-5模型
- 卷积层
- 池化层
- 卷积层
- 池化层
- 全连接层
- 全连接层
全连接层(近似)
代码
train
import osimport numpy as npimport tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_dataimport mnist_inference#data paramsBATCH_SIZE = 100LEARNING_RATE_BASE = 0.8LEARNING_RATE_DECAY = 0.99REGULARAZTION_RATE = 0.0001TRAINING_STEPS = 20000MOVING_AVERAGE_DECAY = 0.99#model save path and nameMODEL_SAVE_PATH="/path/to/model"MODEL_NAME = "model.ckpt"def train(mnist): x = tf.placeholder(tf.float32,[BATCH_SIZE,mnist_inference.IMAGE_SIZE, mnist_inference.IMAGE_SIZE,mnist_inference.NUM_CHANNELS],name='x-input') y_ = tf.placeholder(tf.float32,[None,mnist_inference.OUTPUT_NODE],name='y-input') regularizer = tf.contrib.layers.l2_regularizer(REGULARAZTION_RATE) y = mnist_inference.inference(x,1,regularizer) #step to control the delay global_step = tf.Variable(0,trainable=False) variable_averages = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY,global_step) variable_averages_op = variable_averages.apply(tf.trainable_variables()) #cross entropy and add the regularization # cross_entropy = -tf.reduce_sum(y_*tf.log(y)) cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y,labels=tf.argmax(y_,1)) cross_entropy_mean = tf.reduce_mean(cross_entropy) loss = cross_entropy_mean + tf.add_n(tf.get_collection('losses')) learning_rate = tf.train.exponential_decay(LEARNING_RATE_BASE,global_step, mnist.train.num_examples/BATCH_SIZE, LEARNING_RATE_DECAY) # tf.scalar_summary('learning_rate', learning_rate) train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss,global_step=global_step) #updata the W and variable average at the same time with tf.control_dependencies([train_step,variable_averages_op]): train_op = tf.no_op(name='train') #save the model saver = tf.train.Saver() with tf.Session() as sess: sess.run(tf.global_variables_initializer()) for i in range(TRAINING_STEPS): xs,ys = mnist.train.next_batch(BATCH_SIZE) reshaped_xs = np.reshape(xs,(BATCH_SIZE,mnist_inference.IMAGE_SIZE, mnist_inference.IMAGE_SIZE,mnist_inference.NUM_CHANNELS)) _, loss_value,step = sess.run([train_op, loss,global_step], feed_dict={x: reshaped_xs, y_: ys}) if i % 1000 == 0: # print "step %d, training accuracy %g" % (i, train_accuracy) print ("step %d,loss is %g" % (step,loss_value)) saver.save(sess,os.path.join(MODEL_SAVE_PATH,MODEL_NAME),global_step=global_step)def main(argv=None): mnist = input_data.read_data_sets("/path/to/MNIST_data/", one_hot=True) train(mnist)if __name__=='__main__': tf.app.run()
inference
import tensorflow as tfimport numpy as npfrom tensorflow.examples.tutorials.mnist import input_data#dataset paramsINPUT_NODE = 784OUTPUT_NODE = 10#cnns paramsIMAGE_SIZE = 28NUM_CHANNELS =1NUM_LABELS =10DROP_PROB = 0.5CON1_DEEP = 32CON1_SIZE = 5CON2_DEEP = 64CON2_SIZE = 5FC1_SIZE = 1024#inference structuredef inference(input_tensor,train,regularizer): #first layer with tf.variable_scope('layer1-conv1'): conv1_W = tf.get_variable("weight",[CON1_SIZE,CON1_SIZE,NUM_CHANNELS,CON1_DEEP] ,initializer=tf.truncated_normal_initializer(stddev=0.1)) conv1_b = tf.get_variable("bias",[CON1_DEEP],initializer=tf.constant_initializer(0.1)) #5 * 5 patch ,step 1 ,fill 0 conv1 = tf.nn.conv2d(input_tensor,conv1_W,strides=[1,1,1,1],padding='SAME') relu1 = tf.nn.relu(tf.nn.bias_add(conv1,conv1_b)) with tf.variable_scope('layer1-max_pool'): pool1 = tf.nn.max_pool(relu1,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME') #second layer with tf.variable_scope('layer2-conv2'): conv2_W = tf.get_variable("weight",[CON2_SIZE,CON2_SIZE,CON1_DEEP,CON2_DEEP], initializer=tf.truncated_normal_initializer(stddev=0.1)) conv2_b = tf.get_variable("bias",[CON2_DEEP],initializer=tf.constant_initializer(0.1)) #5*5 patch,step 1 ,fill 0 conv2 = tf.nn.conv2d(pool1,conv2_W,strides=[1,1,1,1],padding='SAME') relu2 = tf.nn.relu(tf.nn.bias_add(conv2,conv2_b)) with tf.variable_scope('layer2-max_pool'): pool2 = tf.nn.max_pool(relu2,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME') #fc1 layer pool_shape = pool2.get_shape().as_list() nodes = pool_shape[1] * pool_shape[2] * pool_shape[3] reshaped = tf.reshape(pool2,[-1,nodes]) with tf.variable_scope('layer3-fc1'): fc1_W = tf.get_variable("weight",[nodes,FC1_SIZE], initializer=tf.truncated_normal_initializer(stddev=0.1)) #regularizer if regularizer != None: tf.add_to_collection('losses',regularizer(fc1_W)) fc1_b = tf.get_variable("bias",[FC1_SIZE],initializer=tf.constant_initializer(0.1)) fc1 = tf.nn.relu(tf.matmul(reshaped,fc1_W)+fc1_b) if train: fc1 = tf.nn.dropout(fc1,0.5) with tf.variable_scope('layer3-softmax'): fc2_W = tf.get_variable("weight",[FC1_SIZE,NUM_LABELS], initializer=tf.truncated_normal_initializer(stddev=0.1)) if regularizer != None: tf.add_to_collection('losses',regularizer(fc2_W)) fc2_b = tf.get_variable("bias",[NUM_LABELS],initializer=tf.constant_initializer(0.1)) y_conv = tf.nn.softmax(tf.matmul(fc1,fc2_W)+fc2_b) return y_conv
eval
import timeimport tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_dataimport numpy as npimport mnist_inferenceimport mnist_trainEVAL_INTERVAL_SECS = 10def evaluate(mnist): with tf.Graph().as_default() as g: x = tf.placeholder(tf.float32, [None, mnist_inference.IMAGE_SIZE, mnist_inference.IMAGE_SIZE, mnist_inference.NUM_CHANNELS], name='x-input') y_ = tf.placeholder(tf.float32,[None,mnist_inference.OUTPUT_NODE], name='y-input') xs = mnist.test.images reshaped_xs = np.reshape(xs, (-1, mnist_inference.IMAGE_SIZE, mnist_inference.IMAGE_SIZE, mnist_inference.NUM_CHANNELS)) test_feed = {x:reshaped_xs,y_:mnist.test.labels} y = mnist_inference.inference(x,None,None) correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(y_,1)) accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32)) variable_averages = tf.train.ExponentialMovingAverage(mnist_train.MOVING_AVERAGE_DECAY) variable_to_restore = variable_averages.variables_to_restore() saver = tf.train.Saver(variable_to_restore) while True: with tf.Session() as sess: ckpt = tf.train.get_checkpoint_state(mnist_train.MODEL_SAVE_PATH) if ckpt and ckpt.model_checkpoint_path: saver.restore(sess,ckpt.model_checkpoint_path) global_stop = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1] accuracy_score = sess.run(accuracy,feed_dict=test_feed) print ("step %s ,accuracy is %g" %(global_stop,accuracy_score)) else: print ("NOT FOUND FILE") return time.sleep(EVAL_INTERVAL_SECS)def main(argv=None): mnist = input_data.read_data_sets("/path/to/MNIST_data/", one_hot=True) evaluate(mnist)if __name__=='__main__': tf.app.run()
阅读全文
0 0
- 人工智能从入门到精通(15)-卷积网络在数字识别的应用
- 人工智能从入门到精通(1)
- 人工智能从入门到精通(2)
- 人工智能从入门到精通(3)
- 人工智能从入门到精通(4)
- 人工智能从入门到精通(5)
- 人工智能从入门到精通(6)
- 人工智能从入门到精通(9)
- 人工智能从入门到精通(10)
- 人工智能从入门到精通(12)
- 人工智能从入门到精通(13)
- 人工智能从入门到精通(14)
- 人工智能从入门到精通(16)
- 人工智能从入门到精通(17)
- 人工智能从入门到精通(18)
- 神经网络入门(二)卷积网络在图像识别的应用
- 人工智能从入门到精通(7)-mnist准备
- 人工智能从入门到精通(8)-mnist实现
- MySQL中文乱码解决办法
- Oplayer 图形绘制
- tpl 文件
- C++:STL常用模块总结(map)
- 知识学习——Servlet基础
- 人工智能从入门到精通(15)-卷积网络在数字识别的应用
- 事务ACID特性及4种隔离级别详解
- Java之二维数组求平均值
- tomcat中运行war包
- 高斯-赛德尔迭代法
- 从0开始搭建ss多用户控流vps
- 软件使用说明书
- 014大数据课程知识点习小结
- java第一周培训学习心得