from tensorflow.examples.tutorials.mnist import input_data from network import network_mnist import matplotlib.pyplot as plt import numpy as np size_input = 28*28 size_output = 10 model_path = './mnist_model/' mnist = input_data.read_data_sets("../MNIST_data", one_hot=True) x = tf.placeholder(tf.float32, [None, size_input]) #数据 y = tf.placeholder(tf.float32, [None, size_output]) #标签 y_pre = network_mnist(x, size_input, size_output) #预测值,预测标签 # compute the accuracy correct_predictions = tf.equal(tf.argmax(y_pre, 1), tf.argmax(y, 1)) accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32)) saver = tf.train.Saver() sess = tf.Session() init = tf.global_variables_initializer() sess.run(init) saver.restore(sess, model_path) y_pre_get, acc_test = sess.run([y_pre, accuracy], feed_dict={x: mnist.test.images, y: mnist.test.labels}) print('acc_test: %.4f%%'%(acc_test*100)) print('***********DONE***************') fig, ax = plt.subplots(nrows=4, ncols=5,
from network import network_mnist import matplotlib.pyplot as plt import numpy as np size_input = 28 * 28 size_output = 10 model_path = './mnist_model/' mnist = input_data.read_data_sets("../MNIST_data", one_hot=True) x = tf.placeholder(tf.float32, [None, size_input]) #数据 y = tf.placeholder(tf.float32, [None, size_output]) #标签 keep_prob_layer = tf.placeholder(tf.float32) y_pre = network_mnist(x, size_input, size_output, keep_prob_layer) #预测值,预测标签 # compute the accuracy correct_predictions = tf.equal(tf.argmax(y_pre, 1), tf.argmax(y, 1)) accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32)) saver = tf.train.Saver() sess = tf.Session() init = tf.global_variables_initializer() sess.run(init) saver.restore(sess, model_path) y_pre_get, acc_test = sess.run([y_pre, accuracy], feed_dict={ x: mnist.test.images, y: mnist.test.labels, keep_prob_layer: 1
batch_size = 512 step_max = int(55000 / batch_size) acc_stop = 0.99 model_path = './mnist_model/' shutil.rmtree(model_path, ignore_errors=True) os.mkdir(model_path) mnist = input_data.read_data_sets("../MNIST_data", one_hot=True) x = tf.placeholder(tf.float32, [None, size_input * size_input]) # 数据 y = tf.placeholder(tf.float32, [None, size_output]) # 标签 x_image = tf.reshape(x, [-1, size_input, size_input, c_in]) # 转换为图像的格式 y_pre = network_mnist(x_image, c_in, size_output) # 预测值,预测标签 cross_entropy = tf.reduce_mean( tf.reduce_sum(-y * tf.log(y_pre), reduction_indices=[1])) train = tf.train.GradientDescentOptimizer(0.08).minimize(cross_entropy) # compute the accuracy correct_predictions = tf.equal(tf.argmax(y_pre, 1), tf.argmax(y, 1)) accuracy = tf.reduce_mean(tf.cast(correct_predictions, tf.float32)) saver = tf.train.Saver() sess = tf.Session() init = tf.global_variables_initializer() sess.run(init) acc_max = -1 str_time = time.strftime('%H:%M:%S', time.localtime()) print('%s: begin' % (str_time))