tensorflow实现BP模型，保存与重新恢复操作

前言：这种方法保存和重新恢复模型参数，是必须要重新定义BP模型的。

还有一种不需要重新定义模型结果的方法，请先参考其他文章，我做了后再重新贴上了。

 

BP模型的建立和训练：

from __future__ import print_function
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

# Parameters
learning_rate = 0.01
num_steps = 1000
batch_size = 100
display_step = 100

# Network Parameters建立的是五层，可定义节点数的BP神经网络，层数可以自行修改

n_hidden_1 = 10  # 1st layer number of neurons
n_hidden_2 = 10  # 2nd layer number of neurons
n_hidden_3 = 10  # 3rd layer number of neurons
n_hidden_4 = 10  # 4th layer number of neurons
n_hidden_5 = 10  # 5th layer number of neurons
num_input = 2  # data input (img shape: 2)
num_classes = 2  # total classes (2 digits)

# tf Graph input
X = tf.placeholder("float", [None, num_input])
Y = tf.placeholder("float", [None, num_classes])

# Store layers weight & bias
with tf.name_scope('parameters'):
    weights = {
        'h1': tf.Variable(tf.random_normal([num_input, n_hidden_1])),
        'h2': tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2])),
        'h3': tf.Variable(tf.random_normal([n_hidden_2, n_hidden_3])),
        'h4': tf.Variable(tf.random_normal([n_hidden_3, n_hidden_4])),
        'h5': tf.Variable(tf.random_normal([n_hidden_4, n_hidden_5])),
        'out': tf.Variable(tf.random_normal([n_hidden_5, num_classes]))
    }
    biases = {
        'b1': tf.Variable(tf.random_normal([n_hidden_1])),
        'b2': tf.Variable(tf.random_normal([n_hidden_2])),
        'b3': tf.Variable(tf.random_normal([n_hidden_3])),
        'b4': tf.Variable(tf.random_normal([n_hidden_4])),
        'b5': tf.Variable(tf.random_normal([n_hidden_5])),
        'out': tf.Variable(tf.random_normal([num_classes]))
    }


# Create model
def neural_net(x):
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['h1']), biases['b1']))

    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['h2']), biases['b2']))

    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['h3']), biases['b3']))

    layer_4 = tf.nn.sigmoid(tf.add(tf.matmul(layer_3, weights['h4']), biases['b4']))

    layer_5 = tf.nn.sigmoid(tf.add(tf.matmul(layer_4, weights['h5']), biases['b5']))

    out_layer = tf.matmul(layer_5, weights['out']) + biases['out']
    return out_layer


def classifier(x, y):
    #    input the expression you want
    if x ** 2 + y ** 2 < 0.1 or 5 * (x - 1.1) ** 4 < y:#两个**代表乘方y^2
        return True
    return False


# Construct model
logits = neural_net(X)
prediction = tf.nn.sigmoid(logits)
# Define loss and optimizer
loss_op = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=Y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
train_op = optimizer.minimize(loss_op)

# Evaluate model
correct_pred = tf.equal(tf.argmax(prediction, 1), tf.argmax(Y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))

# Initialize the variables (i.e. assign their default value)
init = tf.global_variables_initializer()#初始化参数

#定义保存模型参数句柄
saver_path = './model/BP_model.ckpt'
saver = tf.train.Saver()

loss_data = []
acc_data = []
x_data = range(1, num_steps + 1)

# Start training
with tf.Session() as sess:
    # Run the initializer
    sess.run(init)

    for step in range(1, num_steps + 1):
        batch_x = np.random.random((batch_size, 2))
        batch_y = np.zeros((batch_size, 2))
        for i in range(batch_size):
            if classifier(batch_x[i, 0], batch_x[i, 1]):
                batch_y[i, 0] = 1
            else:
                batch_y[i, 1] = 1
        learning_rate = np.exp(-0.1 * step) * 0.1

        # Run optimization op (backprop)
        train, loss, acc = sess.run([train_op, loss_op, accuracy], feed_dict={X: batch_x, Y: batch_y})
        loss_data.append(loss)
        acc_data.append(acc)

    print("Optimization Finished!")

    #保存模型参数
    saved_path = saver.save(sess, saver_path)  # 这个保存了三个东西， .meta是图的结构， 还有两个是模型中变量的值

    plt.plot(x_data, loss_data)
    plt.show()
    plt.plot(x_data, acc_data)
    plt.show()      #凡是使用matplot的需要使用show函数，才能出现图

    xx = np.linspace(0, 1, 100)
    yy = np.linspace(0, 1, 100)
    ZZ = np.zeros((100, 100))
    XX, YY = np.meshgrid(xx, yy)
    rightsum = 0
    for i in range(100):
        for j in range(100):
            inp = np.zeros((1, 2))
            inp[0, 0] = XX[i, j]
            inp[0, 1] = YY[i, j]
            poss = sess.run(prediction, feed_dict={X: inp})
            if poss[0, 0] > poss[0, 1]:
                ZZ[i, j] = 1
                if classifier(inp[0, 0], inp[0, 1]):
                    rightsum += 1
            else:
                ZZ[i, j] = 0
                if classifier(inp[0, 0], inp[0, 1]) == False:
                    rightsum += 1
    print('Testing Accuracy', rightsum / 100, '%')
    plt.contourf(XX, YY, ZZ, 2, colors=('r', 'w', 'b'))

BP模型的恢复和使用：

#！/usr/bin/env python
# _*_ coding:utf-8 _*_

from __future__ import print_function
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

# Parameters
learning_rate = 0.01
num_steps = 1000
batch_size = 100
display_step = 100

# Network Parameters建立的是五层，可定义节点数的BP神经网络，层数可以自行修改

n_hidden_1 = 10  # 1st layer number of neurons
n_hidden_2 = 10  # 2nd layer number of neurons
n_hidden_3 = 10  # 3rd layer number of neurons
n_hidden_4 = 10  # 4th layer number of neurons
n_hidden_5 = 10  # 5th layer number of neurons
num_input = 2  # data input (img shape: 2)
num_classes = 2  # total classes (2 digits)

# tf Graph input
X = tf.placeholder("float", [None, num_input])
Y = tf.placeholder("float", [None, num_classes])

# Store layers weight & bias
with tf.name_scope('parameters'):
    weights = {
        'h1': tf.Variable(tf.random_normal([num_input, n_hidden_1])),
        'h2': tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2])),
        'h3': tf.Variable(tf.random_normal([n_hidden_2, n_hidden_3])),
        'h4': tf.Variable(tf.random_normal([n_hidden_3, n_hidden_4])),
        'h5': tf.Variable(tf.random_normal([n_hidden_4, n_hidden_5])),
        'out': tf.Variable(tf.random_normal([n_hidden_5, num_classes]))
    }
    biases = {
        'b1': tf.Variable(tf.random_normal([n_hidden_1])),
        'b2': tf.Variable(tf.random_normal([n_hidden_2])),
        'b3': tf.Variable(tf.random_normal([n_hidden_3])),
        'b4': tf.Variable(tf.random_normal([n_hidden_4])),
        'b5': tf.Variable(tf.random_normal([n_hidden_5])),
        'out': tf.Variable(tf.random_normal([num_classes]))
    }


# Create model
def neural_net(x):
    layer_1 = tf.nn.sigmoid(tf.add(tf.matmul(x, weights['h1']), biases['b1']))

    layer_2 = tf.nn.sigmoid(tf.add(tf.matmul(layer_1, weights['h2']), biases['b2']))

    layer_3 = tf.nn.sigmoid(tf.add(tf.matmul(layer_2, weights['h3']), biases['b3']))

    layer_4 = tf.nn.sigmoid(tf.add(tf.matmul(layer_3, weights['h4']), biases['b4']))

    layer_5 = tf.nn.sigmoid(tf.add(tf.matmul(layer_4, weights['h5']), biases['b5']))

    out_layer = tf.matmul(layer_5, weights['out']) + biases['out']
    return out_layer

# Construct model
logits = neural_net(X)
prediction = tf.nn.sigmoid(logits)
# Define loss and optimizer
loss_op = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits, labels=Y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate)
train_op = optimizer.minimize(loss_op)

# Evaluate model
correct_pred = tf.equal(tf.argmax(prediction, 1), tf.argmax(Y, 1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))

# Initialize the variables (i.e. assign their default value)
init = tf.global_variables_initializer()#初始化参数

meta_path = './model/BP_model.ckpt.meta'
model_path = './model/BP_model.ckpt'
# saver = tf.train.import_meta_graph(meta_path) # 导入图
saver = tf.train.Saver()

with tf.Session() as sess:
    saver.restore(sess, model_path)  # 导入变量值
    # graph = tf.get_default_graph()
    # sess.run(init)
    t1 = np.zeros((1, 2))
    t1[0, 0] = 1.0
    t1[0, 1] = 2.0
    print(sess.run(prediction, feed_dict={X: t1}))