tensorflow 学习笔记

了解学习下近期热门的机器学习相关基础知识。

开发环境

在windows安装Anaconda以后切换版本python 3.5以后安装tensorflow一直失败,宝宝也是很委屈..
以下详细操作步骤可以看上边链接得视频哦

下载VMware、Ubuntu安装

安装python 2.7、pip 安装python 2.7、pip

1 2	sudo apt install python sudo apt install python-pip

安装tensorflow 安装tensorflow
- TensorFlow 镜像 TensorFlow 镜像

检查

1
2
3

>>> python
>>> import tensorflow as tf
>>> tf.__version__

VMware相关

打开虚拟机
菜单> 虚拟机 > 安装VMware-Tools
点击桌面出现得光驱
复制其中的.tar.gz文件到其他文件夹,解压
1
tar zxpf xxxx.tar.gz
进入解压后的文件夹
1
sudo ./vmware-install.pl
一路回车默认选择后成功安装
1
-- the VMware team
菜单> 查看 > 自动调整大小

mnist 数字识别

下载mnist放入data文件夹

安装PIL

1	pip install -i https://pypi.douban.com/simple pillow

写代码

前向传播

#coding=utf-8
import tensorflow as tf
# 定义输入 & 输出 & 隐藏层
InNode = 784
OutNode = 10
LayNode = 500
# regularizer 正则化
# 随机生成w
def get_weight(shape,regularizer):
    w = tf.Variable(tf.truncated_normal(shape,stddev=0.1))
    if regularizer != None: tf.add_to_collection('losses',tf.contrib.layers.l2_regularizer(regularizer)(w))
    return w
# get
def get_bias(shape):
    b = tf.Variable(tf.zeros(shape))
    return b
# 搭建神经元网络
def forward(x,regularizer):
    w1 = get_weight([InNode,LayNode],regularizer)
    b1 = get_bias([LayNode])
    y1 = tf.nn.relu(tf.matmul(x,w1) + b1)

    w2 = get_weight([LayNode,OutNode],regularizer)
    b2 = get_bias([OutNode])
    y2 = tf.matmul(y1,w2) + b2
    return y2

反向传播

#coding=utf-8
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import mnist_forward
import os

# 输入量，学习率，衰减率，正则化系数，总轮数，滑动衰减，保存路径，保存名
BatchSize = 200
LearningBase = 0.1
LearningDecay = 0.99
Regularizer = 0.0001
Steps = 50000
MovingDecay = 0.99
ModelPath = './model/'
ModelName = 'mnist_model'
# 反向
def backward(mnist):
    x = tf.placeholder(tf.float32,[None, mnist_forward.InNode])
    y_ = tf.placeholder(tf.float32,[None, mnist_forward.OutNode])
    y = mnist_forward.forward(x,Regularizer)
    global_step = tf.Variable(0,trainable=False)

    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'))

    learningRate = tf.train.exponential_decay(
        LearningBase,
        global_step,
        mnist.train.num_examples / BatchSize,
        LearningDecay,
        staircase=True
    )

    train_step = tf.train.GradientDescentOptimizer(learningRate).minimize(loss,global_step=global_step)
    ema = tf.train.ExponentialMovingAverage(MovingDecay,global_step)
    ema_op = ema.apply(tf.trainable_variables())
    with tf.control_dependencies([train_step,ema_op]):
        train_op = tf.no_op(name='train')
    
    saver = tf.train.Saver()

    with tf.Session() as sess:
        init_op = tf.global_variables_initializer()
        sess.run(init_op)
        # 断点续训
        ckpt = tf.train.get_checkpoint_state(ModelPath)
        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(BatchSize)
            _, loss_value, step = sess.run([train_op,loss,global_step],feed_dict={x: xs,y_:ys})
            if i % 1000 == 0:
                print "%d steps,loss: %g" %(step,loss_value)
                saver.save(sess,os.path.join(ModelPath,ModelName),global_step=global_step)

def main():
    mnist = input_data.read_data_sets('./data/',one_hot=True)
    backward(mnist)
if __name__ == '__main__':
    main()

测试

#coding=utf-8
import time
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import mnist_forward
import mnist_backward
TestSecs = 5

# 测试

def test(mnist):
    with tf.Graph().as_default() as g:
        # 定义x,y
        x = tf.placeholder(tf.float32,[None,mnist_forward.InNode])
        y_ = tf.placeholder(tf.float32,[None,mnist_forward.OutNode])
        y = mnist_forward.forward(x,None)

        ema = tf.train.ExponentialMovingAverage(mnist_backward.MovingDecay)
        ema_restore = ema.variables_to_restore()
        # 实例化saver
        saver = tf.train.Saver(ema_restore)
        # 计算正确率
        correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(y_,1))
        accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))

        while True:
            with tf.Session() as sess:
                # 加载模型
                ckpt = tf.train.get_checkpoint_state(mnist_backward.ModelPath)
                if ckpt and ckpt.model_checkpoint_path:
                    saver.restore(sess,ckpt.model_checkpoint_path)
                    global_step = ckpt.model_checkpoint_path.split("/")[-1].split("-")[-1]
                    tempDict = {
                        x: mnist.test.images,
                        y_: mnist.test.labels
                    }
                    accuracy_score = sess.run(accuracy,feed_dict=tempDict)
                    print "%s steps,acc: %g" %(global_step,accuracy_score)
                else:
                    print  "not ckpt"
                    return
            time.sleep(TestSecs)
def main():
    mnist = input_data.read_data_sets('./data/',one_hot=True)
    test(mnist)
if __name__ == '__main__':
    main()

图片识别

#coding=utf-8
import tensorflow as tf
import numpy as np
from PIL import Image
import mnist_backward
import mnist_forward

def restore_model(testPicArr):
    with tf.Graph().as_default() as tg:
        x = tf.placeholder(tf.float32,[None,mnist_forward.InNode])
        y = mnist_forward.forward(x,None)
        preValue = tf.argmax(y,1)

        variable_averages = tf.train.ExponentialMovingAverage(mnist_backward.MovingDecay)
        variable_restore = variable_averages.variables_to_restore()
        saver = tf.train.Saver(variable_restore)

        with tf.Session() as sess:
            # 断点续训
            ckpt = tf.train.get_checkpoint_state(mnist_backward.ModelPath)
            if ckpt and ckpt.model_checkpoint_path:
                saver.restore(sess,ckpt.model_checkpoint_path)
                preValue = sess.run(preValue,feed_dict={x: testPicArr})
                return preValue
            else:
                print "not ckpt !"
                return -1    

def pre_pic(picName):
    img = Image.open(picName)
    reIm = img.resize((28,28),Image.ANTIALIAS)
    im_arr = np.array(reIm.convert('L'))
    threshold = 50
    for i in range(28):
        for j in range(28):
            im_arr[i][j] = 255 - im_arr[i][j]
            if(im_arr[i][j] < threshold):
                im_arr[i][j] = 0
            else: im_arr[i][j] = 255
    
    nm_arr = im_arr.reshape([1,784])
    nm_arr = nm_arr.astype(np.float32)
    img_ready = np.multiply(nm_arr,1.0/255.0)
    return img_ready

def application():
    testPic = raw_input("img of :")
    testPicArr = pre_pic(testPic)
    preValue = restore_model(testPicArr)
    print "the number is:",preValue
def main():
    application()
if __name__  == '__main__':
    main()

相关链接

开发环境

VMware相关

mnist 数字识别

前向传播

反向传播

测试

图片识别

zhangyuhan2016