锐客网

  1. 首页 > it技术 > >

卷积神经网络2完全解析

#coding:utf-8
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import mnist_lenet5_forward
import os
import numpy as np
BATCH_SIZE = 100#一次喂入神经网络图片数量
LEARNING_RATE_BASE =0.005 #学习率0.005
LEARNING_RATE_DECAY = 0.99 #学习衰减率
REGULARIZER = 0.0001
STEPS = 50000 #迭代次数
MOVING_AVERAGE_DECAY = 0.99 #滑动平均衰减率
MODEL_SAVE_PATH="./model/" #模块路径
MODEL_NAME="mnist_model" #模块名称
def backward(mnist):
x = tf.placeholder(tf.float32,[#浮点型
BATCH_SIZE,#喂入图片数量
mnist_lenet5_forward.IMAGE_SIZE,#行分辨率
mnist_lenet5_forward.IMAGE_SIZE,#列分辨率
mnist_lenet5_forward.NUM_CHANNELS]) #通道数
y_ = tf.placeholder(tf.float32, [None, mnist_lenet5_forward.OUTPUT_NODE])
y = mnist_lenet5_forward.forward(x,True, REGULARIZER) #调用向前传播过程
global_step = tf.Variable(0, trainable=False) #全局计数器初始化为零
#交叉熵
ce = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y, labels=tf.argmax(y_, 1))
cem = tf.reduce_mean(ce)
loss = cem + tf.add_n(tf.get_collection('losses'))#对得到的向量求均值
#指数衰减学习率
learning_rate = tf.train.exponential_decay(
LEARNING_RATE_BASE,#学习率0.005
global_step,
mnist.train.num_examples / BATCH_SIZE,
LEARNING_RATE_DECAY,#0.99 #学习衰减率
staircase=True) #阶梯衰减
#梯度下降算法
train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss, global_step=global_step)
#滑动平均模型
ema = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)
ema_op = ema.apply(tf.trainable_variables())
with tf.control_dependencies([train_step, ema_op]):
train_op = tf.no_op(name='train')#将train_step和ema_op绑定到train_op
saver = tf.train.Saver() #实例化一个保存和恢复变量saver
with tf.Session() as sess:
init_op = tf.global_variables_initializer()
sess.run(init_op)
ckpt = tf.train.get_checkpoint_state(MODEL_SAVE_PATH)
if ckpt and ckpt.model_checkpoint_path:
saver.restore(sess, ckpt.model_checkpoint_path)
for i in range(STEPS):
xs, ys = mnist.train.next_batch(BATCH_SIZE) #读取100数据
reshaped_xs = np.reshape(xs,(#转换成相同矩阵
BATCH_SIZE,#100
mnist_lenet5_forward.IMAGE_SIZE,#行分辨率
mnist_lenet5_forward.IMAGE_SIZE,#列分辨率
mnist_lenet5_forward.NUM_CHANNELS))#通道
_, loss_value, step = sess.run([train_op, loss, global_step], feed_dict={x: reshaped_xs, y_: ys})
if i % 100 == 0:
print("After %d training step(s), loss on training batch is %g." % (step, loss_value))
saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME), global_step=global_step)
def main():
mnist = input_data.read_data_sets("./data/", one_hot=True)
backward(mnist)
if __name__ == '__main__':
【卷积神经网络2完全解析】main()

    推荐阅读


    上一篇：小沙盘大世界～沙盘游戏

    下一篇：人已去，城仍在。