Word2Vec实现文本识别分类

深度学习训练营之使用Word2Vec实现文本识别分类

  • 原文链接
  • 环境介绍
  • 前言
  • 前置工作
    • 设置GPU
    • 数据查看
    • 构建数据迭代器
  • Word2Vec的调用
  • 生成数据批次和迭代器
  • 模型训练
    • 初始化
    • 拆分数据集并进行训练
  • 预测

原文链接

  • 本文为365天深度学习训练营 中的学习记录博客
  • 参考文章:365天深度学习训练营-第N4周:用Word2Vec实现文本分类
  • 原作者:K同学啊|接辅导、项目定制

环境介绍

  • 语言环境:Python3.9.12
  • 编译器:jupyter notebook
  • 深度学习环境:TensorFlow2

前言

本次内容我本来是使用miniconda的环境的,但是好像有文件发生了损坏,出现了如下报错,据我所了解应该是某个文件发生了损坏,应该是之前将anaconda误删有关,有所了解或者有同样问题的朋友可以一起进行探讨

前置工作

设置GPU

如果

# 先进行数据加载
import torch
import torch.nn as nn
import torchvision
import os,PIL,pathlib,warnings
import time
from torchvision import transforms, datasets
from torch import nn
from torch.utils.data.dataset import random_split

warnings.filterwarnings("ignore")#忽略警告信息
device=torch.device("cuda"if torch.cuda.is_available()else "cpu")
device

device(type=‘cpu’)

数据查看

本次使用的数据集和之前中文文本识别分类的是一样的

import pandas as pd
train_data=pd.read_csv('train.csv',sep='\t',header=None)
train_data.head()

Word2Vec实现文本识别分类_第1张图片

构建数据迭代器

#构建数据集迭代器
def coustom_data_iter(texts,labels):
    for x,y in zip(texts,labels):
        yield x,y

x=train_data[0].values[:]
y=train_data[1].values[:]    

添加数据迭代器是为了让数据的随机性增强,进行数据集的划分,可以有效的发挥内存的高利用率

Word2Vec的调用

对Word2Vec进行直接的调用

from gensim.models.word2vec import Word2Vec
import numpy as np
#训练浅层神经网络模型
w2v=Word2Vec(vector_size=100,
             min_count=3)

w2v.build_vocab(x)
w2v.train(x,
          total_examples=w2v.corpus_count,
          epochs=30)

build_vocab统计输入每一个词汇出现的次数

def average_vec(text):
    vec=np.zeros(100).reshape((1,100))#表示平均向量
    #(n,100),其中n表示x中的元素的数量 
    for word in text:
        try:
            vec+=w2v.wv[word].reshape((1,100))
        except KeyError:
            continue#未找到,再进行迭代下一个词
    return vec

x_vec=np.concatenate([average_vec(z) for z in x])
w2v.save('w2v_model.pkl')

该步骤将输入的文本转变成了平均向量
对于输入进来的text当中的每一个单词都进行一个查询,确认是否当中有该词,如果有那么就将其添加到vector当中,否则跳出本层循环,查找下一个词.
最后通过np当中的concatenate方法进行一个向量的连接

train_iter=coustom_data_iter(x_vec,y)#训练迭代器
print(len(x),len(y))

12100 12100

设置训练的迭代器

label_name=list(set(train_data[1].values[:]))
print(label_name)
['FilmTele-Play', 'Weather-Query', 'Audio-Play', 'Radio-Listen', 'HomeAppliance-Control', 'Alarm-Update', 'Travel-Query', 'Video-Play', 'Calendar-Query', 'TVProgram-Play', 'Music-Play', 'Other']

生成数据批次和迭代器

text_pipeline=lambda x:average_vec(x)
label_pipeline=lambda x:label_name.index(x)
#lambda语法:lambda  arguments
text_pipeline("我想你了")

Word2Vec实现文本识别分类_第2张图片

label_pipeline("Travel-Query")

6

这里的结果每次都会不太一样,具有一定的随机性

from torch.utils.data import DataLoader

def collate_batch(batch):
    label_list, text_list= [], []
    
    for (_text,_label) in batch:
        # 标签列表
        label_list.append(label_pipeline(_label))
        
        # 文本列表
        processed_text = torch.tensor(text_pipeline(_text), dtype=torch.float32)
        text_list.append(processed_text)
        
        # 偏移量,即语句的总词汇量
        
    label_list = torch.tensor(label_list, dtype=torch.int64)
    text_list  = torch.cat(text_list)
    
    return text_list.to(device),label_list.to(device)

# 数据加载器,调用示例
dataloader = DataLoader(train_iter,
                        batch_size=8,
                        shuffle   =False,
                        collate_fn=collate_batch)

和之前的不同在于没有了offset

模型训练

from torch import nn

class TextClassificationModel(nn.Module):

    def __init__(self, num_class):
        super(TextClassificationModel, self).__init__()
        self.fc = nn.Linear(100, num_class)
    def forward(self, text):
        return self.fc(text)

初始化

num_class  = len(label_name)
vocab_size = 100000
em_size    = 12
model      = TextClassificationModel(num_class).to(device)
import time

def train(dataloader):
    model.train()  # 切换为训练模式
    total_acc, train_loss, total_count = 0, 0, 0
    log_interval = 50
    start_time   = time.time()

    for idx, (text,label) in enumerate(dataloader):
        
        predicted_label = model(text)
        
        optimizer.zero_grad()                    # grad属性归零
        loss = criterion(predicted_label, label) # 计算网络输出和真实值之间的差距,label为真实值
        loss.backward()                          # 反向传播
        torch.nn.utils.clip_grad_norm_(model.parameters(), 0.1) # 梯度裁剪
        optimizer.step()  # 每一步自动更新
        
        # 记录acc与loss
        total_acc   += (predicted_label.argmax(1) == label).sum().item()
        train_loss  += loss.item()
        total_count += label.size(0)
        
        if idx % log_interval == 0 and idx > 0:
            elapsed = time.time() - start_time
            print('| epoch {:1d} | {:4d}/{:4d} batches '
                  '| train_acc {:4.3f} train_loss {:4.5f}'.format(epoch, idx, len(dataloader),
                                              total_acc/total_count, train_loss/total_count))
            total_acc, train_loss, total_count = 0, 0, 0
            start_time = time.time()

def evaluate(dataloader):
    model.eval()  # 切换为测试模式
    total_acc, train_loss, total_count = 0, 0, 0

    with torch.no_grad():
        for idx, (text,label) in enumerate(dataloader):
            predicted_label = model(text)
            
            loss = criterion(predicted_label, label)  # 计算loss值
            # 记录测试数据
            total_acc   += (predicted_label.argmax(1) == label).sum().item()
            train_loss  += loss.item()
            total_count += label.size(0)
            
    return total_acc/total_count, train_loss/total_count

拆分数据集并进行训练

from torch.utils.data.dataset import random_split
from torchtext.data.functional import to_map_style_dataset
# 超参数
EPOCHS     = 30 # epoch
LR         = 5  # 学习率
BATCH_SIZE = 64 # batch size for training

criterion = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=LR)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, 1.0, gamma=0.1)
total_accu = None

# 构建数据集
train_iter = coustom_data_iter(train_data[0].values[:], train_data[1].values[:])
train_dataset = to_map_style_dataset(train_iter)

split_train_, split_valid_ = random_split(train_dataset,
                                          [int(len(train_dataset)*0.8),int(len(train_dataset)*0.2)])

train_dataloader = DataLoader(split_train_, batch_size=BATCH_SIZE,
                              shuffle=True, collate_fn=collate_batch)

valid_dataloader = DataLoader(split_valid_, batch_size=BATCH_SIZE,
                              shuffle=True, collate_fn=collate_batch)

for epoch in range(1, EPOCHS + 1):
    epoch_start_time = time.time()
    train(train_dataloader)
    val_acc, val_loss = evaluate(valid_dataloader)
    
    # 获取当前的学习率
    lr = optimizer.state_dict()['param_groups'][0]['lr']
    
    if total_accu is not None and total_accu > val_acc:
        scheduler.step()
    else:
        total_accu = val_acc
    print('-' * 69)
    print('| epoch {:1d} | time: {:4.2f}s | '
          'valid_acc {:4.3f} valid_loss {:4.3f} | lr {:4.6f}'.format(epoch,
                                           time.time() - epoch_start_time,
                                           val_acc,val_loss,lr))

    print('-' * 69)
| epoch 1 |   50/ 152 batches | train_acc 0.742 train_loss 0.02635
| epoch 1 |  100/ 152 batches | train_acc 0.820 train_loss 0.02033
| epoch 1 |  150/ 152 batches | train_acc 0.838 train_loss 0.01927
---------------------------------------------------------------------
| epoch 1 | time: 0.95s | valid_acc 0.819 valid_loss 0.023 | lr 5.000000
---------------------------------------------------------------------
| epoch 2 |   50/ 152 batches | train_acc 0.850 train_loss 0.01876
| epoch 2 |  100/ 152 batches | train_acc 0.849 train_loss 0.02012
| epoch 2 |  150/ 152 batches | train_acc 0.847 train_loss 0.01736
---------------------------------------------------------------------
| epoch 2 | time: 0.92s | valid_acc 0.869 valid_loss 0.016 | lr 5.000000
---------------------------------------------------------------------
| epoch 3 |   50/ 152 batches | train_acc 0.858 train_loss 0.01588
| epoch 3 |  100/ 152 batches | train_acc 0.833 train_loss 0.02008
| epoch 3 |  150/ 152 batches | train_acc 0.864 train_loss 0.01813
---------------------------------------------------------------------
| epoch 3 | time: 0.86s | valid_acc 0.835 valid_loss 0.023 | lr 5.000000
---------------------------------------------------------------------
| epoch 4 |   50/ 152 batches | train_acc 0.883 train_loss 0.01309
| epoch 4 |  100/ 152 batches | train_acc 0.899 train_loss 0.00996
| epoch 4 |  150/ 152 batches | train_acc 0.895 train_loss 0.00927
---------------------------------------------------------------------
| epoch 4 | time: 0.87s | valid_acc 0.888 valid_loss 0.011 | lr 0.500000
---------------------------------------------------------------------
| epoch 5 |   50/ 152 batches | train_acc 0.906 train_loss 0.00834
...
| epoch 30 |  150/ 152 batches | train_acc 0.900 train_loss 0.00717
---------------------------------------------------------------------
| epoch 30 | time: 0.92s | valid_acc 0.886 valid_loss 0.010 | lr 0.000000
---------------------------------------------------------------------
test_acc, test_loss = evaluate(valid_dataloader)
print('test accuracy {:8.3f}'.format(test_acc))

在这里插入图片描述

预测

def predict(text, text_pipeline):
    with torch.no_grad():
        text = torch.tensor(text_pipeline(text),dtype=torch.float32)
        print(text.shape)
        output = model(text)
        return output.argmax(1).item()
    
ex_text_str = "随便播放一首专辑阁楼里的佛里的歌"
#ex_text_str = "还有双鸭山到淮阴的汽车票吗13号的"
model = model.to(device)

print("该文本的类别是:%s" % label_name[predict(ex_text_str, text_pipeline)])
torch.Size([1, 100])
该文本的类别是:Music-Play

你可能感兴趣的:(深度学习,word2vec,分类,人工智能)