卷积神经网络 CNN

1、feature_map=[（原图片尺寸 - 卷积核尺寸）/ 步长] + 1

• 卷积核

1. 卷积运算是两个卷积核大小的矩阵的内积运算
2. 卷积核的公式有很多，卷积核也称为滤波器：均值滤波器、高斯滤波器、拉普拉斯滤波器
3. 每一层的卷积核大小和个数可以自己定义，不过一般情况下，根据实验得到的经验来看，会在越靠近输入层的卷积层设定少量的卷积核，越往后，卷积层设定的卷积核数目就越多

• 池化层

     池化层方法：MaxPooling(取滑动窗口里最大的值)、AveragePooling(取滑动窗口内所有值的平均值)

     要想经过卷积或池化输出的图片的大小不变，通过zero padding 补零

     补零后的 feature_map=(width + 2 * padding_size - filter_size) / stride +1         

     其中，padding_size 补的尺寸，filter_size 卷积核的尺寸

• Flatten 层 & Fully Connected Layer

     

---

train.py

#! /usr/bin/env python

import tensorflow as tf
import numpy as np
import os
import time
import datetime
import data_helpers
from text_cnn import TextCNN
from tensorflow.contrib import learn

# Parameters
# ==================================================

# Data loading params   语料文件路径定义
tf.flags.DEFINE_float("dev_sample_percentage", .1, "Percentage of the training data to use for validation")
tf.flags.DEFINE_string("positive_data_file", "./data/rt-polaritydata/rt-polarity.pos", "Data source for the positive data.")
tf.flags.DEFINE_string("negative_data_file", "./data/rt-polaritydata/rt-polarity.neg", "Data source for the negative data.")

# Model Hyperparameters  定义网络超参数
tf.flags.DEFINE_integer("embedding_dim", 128, "Dimensionality of character embedding (default: 128)")
tf.flags.DEFINE_string("filter_sizes", "3,4,5", "Comma-separated filter sizes (default: '3,4,5')")
tf.flags.DEFINE_integer("num_filters", 128, "Number of filters per filter size (default: 128)")
tf.flags.DEFINE_float("dropout_keep_prob", 0.5, "Dropout keep probability (default: 0.5)")
tf.flags.DEFINE_float("l2_reg_lambda", 0.0, "L2 regularization lambda (default: 0.0)")

# Training parameters   训练参数
tf.flags.DEFINE_integer("batch_size", 64, "Batch Size (default: 64)")
tf.flags.DEFINE_integer("num_epochs", 200, "Number of training epochs (default: 200)")   # 总训练次数
tf.flags.DEFINE_integer("evaluate_every", 100, "Evaluate model on dev set after this many steps (default: 100)")  # 每训练100次测试一下
tf.flags.DEFINE_integer("checkpoint_every", 100, "Save model after this many steps (default: 100)")   # 保存一次模型
tf.flags.DEFINE_integer("num_checkpoints", 5, "Number of checkpoints to store (default: 5)")
# Misc Parameters
tf.flags.DEFINE_boolean("allow_soft_placement", True, "Allow device soft device placement")   # 加上一个布尔类型的参数，要不要自动分配
tf.flags.DEFINE_boolean("log_device_placement", False, "Log placement of ops on devices")   # 加上一个布尔类型的参数，要不要打印日志

#打印一下相关初始参数
FLAGS = tf.flags.FLAGS
# FLAGS._parse_flags()
# print("\nParameters:")
# for attr, value in sorted(FLAGS.__flags.items()):
#     print("{}={}".format(attr.upper(), value))
# print("")

def preprocess():
    # Data Preparation
    # ==================================================

    # Load data
    print("Loading data...")
    x_text, y = data_helpers.load_data_and_labels(FLAGS.positive_data_file, FLAGS.negative_data_file)

    # Build vocabulary
    max_document_length = max([len(x.split(" ")) for x in x_text])    # 计算最长邮件
    vocab_processor = learn.preprocessing.VocabularyProcessor(max_document_length)   # tensorflow提供的工具，将数据填充为最大长度，默认0填充
    x = np.array(list(vocab_processor.fit_transform(x_text)))

    # Randomly shuffle data  数据洗牌
    np.random.seed(10)
    # np.arange生成随机序列
    shuffle_indices = np.random.permutation(np.arange(len(y)))
    x_shuffled = x[shuffle_indices]
    y_shuffled = y[shuffle_indices]

    # 将数据按训练train和测试dev分块
    # Split train/test set
    # TODO: This is very crude, should use cross-validation
    dev_sample_index = -1 * int(FLAGS.dev_sample_percentage * float(len(y)))
    x_train, x_dev = x_shuffled[:dev_sample_index], x_shuffled[dev_sample_index:]
    y_train, y_dev = y_shuffled[:dev_sample_index], y_shuffled[dev_sample_index:]

    del x, y, x_shuffled, y_shuffled

    print("Vocabulary Size: {:d}".format(len(vocab_processor.vocabulary_)))
    print("Train/Dev split: {:d}/{:d}".format(len(y_train), len(y_dev)))   # 打印切分的比例
    return x_train, y_train, vocab_processor, x_dev, y_dev

def train(x_train, y_train, vocab_processor, x_dev, y_dev):
    # Training
    # ==================================================

    with tf.Graph().as_default():
        session_conf = tf.ConfigProto(
          allow_soft_placement=FLAGS.allow_soft_placement,
          log_device_placement=FLAGS.log_device_placement)
        sess = tf.Session(config=session_conf)
        with sess.as_default():
            # 卷积池化网络导入
            cnn = TextCNN(
                sequence_length=x_train.shape[1],
                num_classes=y_train.shape[1],   # 分几类
                vocab_size=len(vocab_processor.vocabulary_),
                embedding_size=FLAGS.embedding_dim,
                filter_sizes=list(map(int, FLAGS.filter_sizes.split(","))),   # 上面定义的filter_sizes拿过来，“3,4,5,”按“，”分割
                num_filters=FLAGS.num_filters,   # 一共有几个filter
                l2_reg_lambda=FLAGS.l2_reg_lambda)  # l2正则化项

            # Define Training procedure
            global_step = tf.Variable(0, name="global_step", trainable=False)
            optimizer = tf.train.AdamOptimizer(1e-3)   # 定义优化器
            grads_and_vars = optimizer.compute_gradients(cnn.loss)
            train_op = optimizer.apply_gradients(grads_and_vars, global_step=global_step)

            # Keep track of gradient values and sparsity (optional)
            grad_summaries = []
            for g, v in grads_and_vars:
                if g is not None:
                    grad_hist_summary = tf.summary.histogram("{}/grad/hist".format(v.name), g)
                    sparsity_summary = tf.summary.scalar("{}/grad/sparsity".format(v.name), tf.nn.zero_fraction(g))
                    grad_summaries.append(grad_hist_summary)
                    grad_summaries.append(sparsity_summary)
            grad_summaries_merged = tf.summary.merge(grad_summaries)

            # Output directory for models and summaries
            timestamp = str(int(time.time()))
            out_dir = os.path.abspath(os.path.join(os.path.curdir, "runs", timestamp))
            print("Writing to {}\n".format(out_dir))

            # Summaries for loss and accuracy   损失函数和准确率的参数保存
            loss_summary = tf.summary.scalar("loss", cnn.loss)
            acc_summary = tf.summary.scalar("accuracy", cnn.accuracy)

            # Train Summaries 训练数据保存
            train_summary_op = tf.summary.merge([loss_summary, acc_summary, grad_summaries_merged])
            train_summary_dir = os.path.join(out_dir, "summaries", "train")
            train_summary_writer = tf.summary.FileWriter(train_summary_dir, sess.graph)

            # Dev summaries  测试数据保存
            dev_summary_op = tf.summary.merge([loss_summary, acc_summary])
            dev_summary_dir = os.path.join(out_dir, "summaries", "dev")
            dev_summary_writer = tf.summary.FileWriter(dev_summary_dir, sess.graph)

            # Checkpoint directory. Tensorflow assumes this directory already exists so we need to create it
            checkpoint_dir = os.path.abspath(os.path.join(out_dir, "checkpoints"))
            checkpoint_prefix = os.path.join(checkpoint_dir, "model")
            if not os.path.exists(checkpoint_dir):
                os.makedirs(checkpoint_dir)
            saver = tf.train.Saver(tf.global_variables(), max_to_keep=FLAGS.num_checkpoints)    # 前面定义好参数num_checkpoints

            # Write vocabulary
            vocab_processor.save(os.path.join(out_dir, "vocab"))

            # Initialize all variables
            sess.run(tf.global_variables_initializer())    # 初始化所有变量

            # 定义训练函数
            def train_step(x_batch, y_batch):
                """
                A single training step
                """
                feed_dict = {
                  cnn.input_x: x_batch,
                  cnn.input_y: y_batch,
                  cnn.dropout_keep_prob: FLAGS.dropout_keep_prob     # 参数在前面有定义
                }
                _, step, summaries, loss, accuracy = sess.run([train_op, global_step, train_summary_op, cnn.loss, cnn.accuracy],feed_dict)
                time_str = datetime.datetime.now().isoformat()      # 取当前时间，python的函数
                print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
                train_summary_writer.add_summary(summaries, step)

            # 定义测试函数
            def dev_step(x_batch, y_batch, writer=None):
                """
                Evaluates model on a dev set
                """
                feed_dict = {
                  cnn.input_x: x_batch,
                  cnn.input_y: y_batch,
                  cnn.dropout_keep_prob: 1.0    # 神经元全部保留
                }
                step, summaries, loss, accuracy = sess.run(
                    [global_step, dev_summary_op, cnn.loss, cnn.accuracy],feed_dict)
                time_str = datetime.datetime.now().isoformat()
                print("{}: step {}, loss {:g}, acc {:g}".format(time_str, step, loss, accuracy))
                if writer:
                    writer.add_summary(summaries, step)

            # Generate batches
            batches = data_helpers.batch_iter(
                list(zip(x_train, y_train)), FLAGS.batch_size, FLAGS.num_epochs)
            # Training loop. For each batch...
            # 训练部分
            for batch in batches:
                x_batch, y_batch = zip(*batch)    # 按batch把数据拿进来
                train_step(x_batch, y_batch)
                current_step = tf.train.global_step(sess, global_step)   # 将Session和global_step值传进来
                if current_step % FLAGS.evaluate_every == 0:      # 每FLAGS.evaluate_every次每100执行一次测试
                    print("\nEvaluation:")
                    dev_step(x_dev, y_dev, writer=dev_summary_writer)
                    print("")
                if current_step % FLAGS.checkpoint_every == 0:    # 每checkpoint_every次执行一次保存模型
                    path = saver.save(sess, checkpoint_prefix, global_step=current_step)  # 定义模型保存路径
                    print("Saved model checkpoint to {}\n".format(path))

def main(argv=None):
    x_train, y_train, vocab_processor, x_dev, y_dev = preprocess()
    train(x_train, y_train, vocab_processor, x_dev, y_dev)

if __name__ == '__main__':
    tf.app.run()

data_helpers:

 

import numpy as np
import re


def clean_str(string):
    """
    Tokenization/string cleaning for all datasets except for SST.
    Original taken from https://github.com/yoonkim/CNN_sentence/blob/master/process_data.py
    """
    # 清理数据替换掉无词义的符号
    string = re.sub(r"[^A-Za-z0-9(),!?\'\`]", " ", string)   # 将字符串中 用空格替换
    string = re.sub(r"\'s", " \'s", string)
    string = re.sub(r"\'ve", " \'ve", string)
    string = re.sub(r"n\'t", " n\'t", string)
    string = re.sub(r"\'re", " \'re", string)
    string = re.sub(r"\'d", " \'d", string)
    string = re.sub(r"\'ll", " \'ll", string)
    string = re.sub(r",", " , ", string)
    string = re.sub(r"!", " ! ", string)
    string = re.sub(r"\(", " \( ", string)
    string = re.sub(r"\)", " \) ", string)
    string = re.sub(r"\?", " \? ", string)
    string = re.sub(r"\s{2,}", " ", string)
    return string.strip().lower()


def load_data_and_labels(positive_data_file, negative_data_file):
    """
    Loads MR polarity data from files, splits the data into words and generates labels.
    Returns split sentences and labels.
    """
    # Load data from files
    positive_examples = list(open(positive_data_file, "r", encoding='utf-8').readlines())  # readlines()自动将文件内容分析成一个行的列表
    positive_examples = [s.strip() for s in positive_examples]   # 去空格   strip()方法只能删除开头或者结尾的字符，不能删除中间部分的字符
    negative_examples = list(open(negative_data_file, "r", encoding='utf-8').readlines())
    negative_examples = [s.strip() for s in negative_examples]
    # Split by words
    x_text = positive_examples + negative_examples
    x_text = [clean_str(sent) for sent in x_text]    # 字符过滤，实现函数见clean_str()
    # Generate labels
    positive_labels = [[0, 1] for _ in positive_examples]
    negative_labels = [[1, 0] for _ in negative_examples]
    y = np.concatenate([positive_labels, negative_labels], 0)   # 将两种label连在一起
    return [x_text, y]


def batch_iter(data, batch_size, num_epochs, shuffle=True):   # shuffle=True洗牌
    """
    Generates a batch iterator for a dataset.
    """
    # 每次只输出shuffled_data[start_index:end_index]这么多
    data = np.array(data)
    data_size = len(data)
    num_batches_per_epoch = int((len(data)-1)/batch_size) + 1  # 每一个epoch有多少个batch_size
    for epoch in range(num_epochs):
        # Shuffle the data at each epoch
        if shuffle:
            shuffle_indices = np.random.permutation(np.arange(data_size))    # 洗牌
            shuffled_data = data[shuffle_indices]
        else:
            shuffled_data = data
        for batch_num in range(num_batches_per_epoch):
            start_index = batch_num * batch_size        # 当前batch的索引
            end_index = min((batch_num + 1) * batch_size, data_size)    # 判断下一个batch是不是超过最后一个数据了
            yield shuffled_data[start_index:end_index]

text_cnn.py

 

import tensorflow as tf
import numpy as np

# 定义CNN网络实现的类
class TextCNN(object):
    """
    A CNN for text classification.
    Uses an embedding layer, followed by a convolutional, max-pooling and softmax layer.
    """
    def __init__(self, sequence_length, num_classes, vocab_size,embedding_size, filter_sizes, num_filters, l2_reg_lambda=0.0):  # 把train.py中TextCNN里定义的参数传进来

        # Placeholders for input, output and dropout
        self.input_x = tf.placeholder(tf.int32, [None, sequence_length], name="input_x")   # input_x输入语料，待训练的内容，维度是sequence_length，“N个词构成的N维向量”
        self.input_y = tf.placeholder(tf.float32, [None, num_classes], name="input_y")   # input_y输入语料，待训练的内容标签，维度是num_classes，“正面||负面”
        self.dropout_keep_prob = tf.placeholder(tf.float32, name="dropout_keep_prob")   # dropout_keep_prob dropout参数，防止过拟合，训练时用

        # Keeping track of l2 regularization loss (optional)
        l2_loss = tf.constant(0.0)

        # Embedding layer
        # 指定运算结构的运行位置在cpu非GPU，因为“embedding”无法运行在GPU   通过tf.name_scope指定“embedding”
        with tf.device('/cpu:0'), tf.name_scope("embedding"):    # 指定cpu
            self.W = tf.Variable(tf.random_uniform([vocab_size, embedding_size], -1.0, 1.0),name="W")   # 定义W 并初始化
            self.embedded_chars = tf.nn.embedding_lookup(self.W, self.input_x)
            self.embedded_chars_expanded = tf.expand_dims(self.embedded_chars, -1)    # 加一个维度，转换为4维的格式

        # Create a convolution + maxpool layer for each filter size
        pooled_outputs = []
        # filter_sizes卷积核尺寸，枚举后遍历
        for i, filter_size in enumerate(filter_sizes):
            with tf.name_scope("conv-maxpool-%s" % filter_size):
                # Convolution Layer
                filter_shape = [filter_size, embedding_size, 1, num_filters]  # 四个参数分别为filter_size高h，embedding_size宽w，channel为1，filter个数
                W = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="W")  # W进行高斯初始化
                b = tf.Variable(tf.constant(0.1, shape=[num_filters]), name="b")    # b给初始化为一个常量
                conv = tf.nn.conv2d(
                    self.embedded_chars_expanded,
                    W,
                    strides=[1, 1, 1, 1],
                    padding="VALID",    # 这里不需要padding
                    name="conv")
                # Apply nonlinearity 激活函数  可以理解为，正面或负面评论有一些标志词汇，这些词汇概率被增强，即一旦出现这些词汇，倾向性分类进正或负面评价，该激励函数可加快学习进度，增加稀疏性，因为让确定的事情更确定，噪声的影响就降到了最低。
                h = tf.nn.relu(tf.nn.bias_add(conv, b), name="relu")
                # Maxpooling over the outputs  池化
                pooled = tf.nn.max_pool(
                    h,
                    ksize=[1, sequence_length - filter_size + 1, 1, 1],
                    strides=[1, 1, 1, 1],
                    padding='VALID',   # 这里不需要padding
                    name="pool")
                pooled_outputs.append(pooled)

        # Combine all the pooled features
        num_filters_total = num_filters * len(filter_sizes)
        self.h_pool = tf.concat(pooled_outputs, 3)
        self.h_pool_flat = tf.reshape(self.h_pool, [-1, num_filters_total])

        # Add dropout
        # drop层，防止过拟合，参数为dropout_keep_prob
        # 过拟合的本质是采样失真，噪声权重影响了判断，如果采样足够多，足够充分，噪声的影响可以被量化到趋近事实，也就无从过拟合
        # 即数据越大，drop和正则化就越不需要
        with tf.name_scope("dropout"):
            self.h_drop = tf.nn.dropout(self.h_pool_flat, self.dropout_keep_prob)

        # Final (unnormalized) scores and predictions   输出层
        with tf.name_scope("output"):
            W = tf.get_variable(
                "W",
                shape=[num_filters_total, num_classes],   # 前面连扁平化后的池化操作
                initializer=tf.contrib.layers.xavier_initializer())   # 定义初始化方式
            b = tf.Variable(tf.constant(0.1, shape=[num_classes]), name="b")
            # 损失函数导入
            l2_loss += tf.nn.l2_loss(W)
            l2_loss += tf.nn.l2_loss(b)
            # xw+b
            self.scores = tf.nn.xw_plus_b(self.h_drop, W, b, name="scores")  # 得分函数
            self.predictions = tf.argmax(self.scores, 1, name="predictions")  # 预测结果

        # Calculate mean cross-entropy loss
        with tf.name_scope("loss"):
            # loss 交叉熵损失函数
            losses = tf.nn.softmax_cross_entropy_with_logits(logits=self.scores, labels=self.input_y)
            self.loss = tf.reduce_mean(losses) + l2_reg_lambda * l2_loss

        # Accuracy
        with tf.name_scope("accuracy"):
            # 准确率，求和计算算数平均值
            correct_predictions = tf.equal(self.predictions, tf.argmax(self.input_y, 1))
            self.accuracy = tf.reduce_mean(tf.cast(correct_predictions, "float"), name="accuracy") 

实验结果：

 

文件夹中生成文件： 

数据源共享网盘：链接：https://pan.baidu.com/s/1JudQBIjUIyrmJYhSdbSEHw   提取码：zpcf 

理解卷积神经网络：Understanding Convolutional Neural Networks for NLP  链接：http://www.wildml.com/2015/11/understanding-convolutional-neural-networks-for-nlp/

还可以参考链接：https://blog.csdn.net/lwplwf/article/details/72723140