深度學習（三）：中英文郵件分類（含代碼）

學習視頻：Tensorflow項目實戰視頻課程-文本分類

主要內容：英文郵件分類、中文郵件分類【中英文郵件分類主要的區別在於embedding處，由於中文需要使用預訓練詞向量，而英文就不需要】

主要方法：TextCnn

一、英文郵件分類的代碼實現

class TextCNN(object):
    def __init__(self, text_length, nclasses, vocab_size,
            embedding_size, filter_sizes, num_filters, l2_reg_lambda=0.0):

        # feeds需要的參數
        self.input_x = tf.placeholder(tf.int32, [None, text_length], name="input_x")
        self.input_y = tf.placeholder(tf.float32, [None, nclasses], name="input_y")
        self.dropout_keep_prob = tf.placeholder(tf.float32, name="dropout_keep_prob")

        l2_loss = tf.constant(0.0)  # 正則懲罰項的w

        # embedding層
        with tf.name_scope('embedding'):
            self.W = tf.Variable(tf.random_uniform([vocab_size, embedding_size], -1.0, 1.0), name="W")
            self.embedded_chars = tf.nn.embedding_lookup(self.W, self.input_x)          
            self.embedded_chars_expanded = tf.expand_dims(self.embedded_chars, -1)      

        # 卷積層和池化層
        pool_outputs = []
        for i, filter_size in enumerate(filter_sizes):  
            with tf.name_scope("conv-maxpool-%s" % filter_size):
                # 卷積層
                filter_shape = [filter_size, embedding_size, 1, num_filters]
                W = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="W")
                b = tf.Variable(tf.constant(0.1, shape=[num_filters]), name="b") 
                conv = tf.nn.conv2d(
                    self.embedded_chars_expanded,
                    W,
                    strides=[1, 1, 1, 1],
                    padding="VALID",
                    name="conv")
                h = tf.nn.relu(tf.nn.bias_add(conv, b), name="relu")    

                # 池化層
                pooled = tf.nn.max_pool(
                    h,
                    ksize=[1, text_length - filter_size + 1, 1, 1],     
                    strides=[1, 1, 1, 1],
                    padding="VALID",
                    name="pool")
                pool_outputs.append(pooled)

        # 全連接層
        num_filters_total = num_filters * len(filter_sizes)     
        self.h_pool = tf.concat(pool_outputs, 0)   
        self.h_pool_flat = tf.reshape(self.h_pool, [-1, num_filters_total])  

        # dropout層
        with tf.name_scope("dropout"):
            self.h_drop = tf.nn.dropout(self.h_pool_flat, self.dropout_keep_prob)

        # 輸出層
        with tf.name_scope("output"):
            W = tf.get_variable(                       
                "W",  
                shape=[num_filters_total, nclasses],  
                initializer=tf.contrib.layers.xavier_initializer()  
            )
            b = tf.Variable(tf.constant(0.1, shape=[nclasses]), name="b")
            l2_loss += tf.nn.l2_loss(W)     # 正則懲罰項值的改變
            l2_loss += tf.nn.l2_loss(b)
            self.fin_out = tf.nn.xw_plus_b(self.h_drop, W, b, name="scores")
            self.predictions = tf.argmax(self.fin_out, 1, name="predictions")       

        # 計算損失
        with tf.name_scope("loss"):
            losses = tf.nn.softmax_cross_entropy_with_logits(logits=self.fin_out, labels=self.input_y)
            self.loss = tf.reduce_mean(losses) + l2_reg_lambda * l2_loss

        # 計算準確度
        with tf.name_scope("accuracy"):
            correct_predictions = tf.equal(self.predictions, tf.argmax(self.input_y, 1))
            self.accuracy = tf.reduce_mean(tf.cast(correct_predictions, "float"), name="accuracy")

二、中文郵件分類的代碼實現

class TextCNN(object):
    def __init__(self, text_length, nclasses, embedding_size, filter_sizes, num_filters, l2_reg_lambda=0.0):

        # feeds需要的參數
        self.input_x = tf.placeholder(tf.float32, [None, text_length, embedding_size], name="input_x")
        self.input_y = tf.placeholder(tf.float32, [None, nclasses], name="input_y")
        self.dropout_keep_prob = tf.placeholder(tf.float32, name="dropout_keep_prob")

        l2_loss = tf.constant(0.0)  # 正則懲罰項的w

        # embedding層
        with tf.name_scope('embedding'):
            self.embedded_chars = self.input_x        
            self.embedded_chars_expanded = tf.expand_dims(self.embedded_chars, -1)   
   
        # 卷積層和池化層
        pool_outputs = []
        for i, filter_size in enumerate(filter_sizes): 
            with tf.name_scope("conv-maxpool-%s" % filter_size):
                # 卷積層
                filter_shape = [filter_size, embedding_size, 1, num_filters]
                W = tf.Variable(tf.truncated_normal(filter_shape, stddev=0.1), name="W")   
                b = tf.Variable(tf.constant(0.1, shape=[num_filters]), name="b")   
                conv = tf.nn.conv2d(
                    self.embedded_chars_expanded,
                    W,
                    strides=[1, 1, 1, 1],
                    padding="VALID",
                    name="conv")
                h = tf.nn.relu(tf.nn.bias_add(conv, b), name="relu")   

                # 池化層
                pooled = tf.nn.max_pool(
                    h,
                    ksize=[1, text_length - filter_size + 1, 1, 1],    
                    strides=[1, 1, 1, 1],
                    padding="VALID",
                    name="pool")
                pool_outputs.append(pooled)

        # 全連接層
        num_filters_total = num_filters * len(filter_sizes)     
        self.h_pool = tf.concat(pool_outputs, 0)   ，
        self.h_pool_flat = tf.reshape(self.h_pool, [-1, num_filters_total])  

        # dropout層
        with tf.name_scope("dropout"):
            self.h_drop = tf.nn.dropout(self.h_pool_flat, self.dropout_keep_prob)

        # 輸出層
        with tf.name_scope("output"):
            W = tf.get_variable(                       
                "W",  
                shape=[num_filters_total, nclasses],  
                initializer=tf.contrib.layers.xavier_initializer()  
            )
            b = tf.Variable(tf.constant(0.1, shape=[nclasses]), name="b")
            l2_loss += tf.nn.l2_loss(W)     # 正則懲罰項值的改變
            l2_loss += tf.nn.l2_loss(b)
            self.fin_out = tf.nn.xw_plus_b(self.h_drop, W, b, name="scores")
            self.predictions = tf.argmax(self.fin_out, 1, name="predictions")      

        # 計算損失
        with tf.name_scope("loss"):
            losses = tf.nn.softmax_cross_entropy_with_logits(logits=self.fin_out, labels=self.input_y)
            self.loss = tf.reduce_mean(losses) + l2_reg_lambda * l2_loss

        # 計算準確度
        with tf.name_scope("accuracy"):
            correct_predictions = tf.equal(self.predictions, tf.argmax(self.input_y, 1))
            self.accuracy = tf.reduce_mean(tf.cast(correct_predictions, "float"), name="accuracy")