Transformer-LSTM-SVM回归

最新推荐文章于 2025-05-01 11:41:26 发布

^ω^宇博

最新推荐文章于 2025-05-01 11:41:26 发布

阅读量768

点赞数 27

分类专栏：数学模型文章标签：机器学习

本文链接：https://blog.csdn.net/2301_81197800/article/details/146563903

版权

数学模型专栏收录该内容

12 篇文章

订阅专栏

题目：Transformer-LSTM-SVM回归

文章目录

题目：Transformer-LSTM-SVM回归
- 前言一：Transformer
- - 1. Transformer的原理
  - - 1.1 Transformer的核心结构
    - 1.2 注意力机制
    - 1.4 位置编码
    - 1.5 损失函数
  - 2. 完整案例
- LSTM
- SVM

前言一：Transformer

1. Transformer的原理

Transformer是一种基于注意力机制的深度学习模型，最早由Vaswani等人在2017年的论文《Attention Is All You Need》中提出。它主要用于序列建模任务，如:自然语言处理和时间序列预测。

1.1 Transformer的核心结构

Transformer主要由两部分组成：

编码器：处理输入数据，将其转化为高维向量表示。
解码器：根据编码的输出和目标序列生成结果。

解码器
包含 $N$ 层，每一层都由两个子层组成：

多头注意力机制。
前向全连接网络。

使用残差连接和层归一化以提高训练效率。

解码器

包含 $N$ 层，每一层都由两个子层组成：

多头注意力机制。
编码器-解码器注意力机制。
前向全连接网络。

使用残差连接和层归一化以提高训练效率。

1.2 注意力机制

注意力得分计算
$\text{Attention}\left(Q,K,V\right)=\text{softmax}\left(\frac{QK^T}{\sqrt{d_k}}V\right)$

$Q$ 是查询向量。
$K$ 键向量。
$V$ 值向量。
$d_k$ :向量维度

多头注意力
多头注意力是将输入分成多组，独立计算每组的注意力，然后拼接在一起：
$\text{MultiHead}\left( Q,K,V\right)=\text{Concat}\left(head_1,head_2,\dots ,head_n\right)W^O$
每个头的计算如下:
$head_i=\text{Attention}\left(QW_{i}^{Q},KW_{i}^{K},VW_{i}^{V}\right)$

$W_{i}^{Q},W_{i}^{K},W_{i}^{V}$ :学习参数矩阵

1.4 位置编码

由于Transformer没有循环结构，如RNN，无法显试捕获序列顺序信息，因此引入位置编码：
$PE\left(pos,2i\right)=\text{sin}\left(\frac{pos}{10000^{\frac{2i}{d}}}\right)$
$PE\left(pos,2i+1\right)=\text{cos}\left(\frac{pos}{10000^{\frac{2i}{d}}}\right)$

$p o s$ :序列中的位置。
$d$ :嵌入向量的维度。

1.5 损失函数

Transformer常常使用交叉熵损失：
$\mathcal{L}=-\frac{1}{N}\sum_{i=1}^{N}\sum_{j=1}^{C}y_{ij}\text{ln}\left(\hat{y_{ij}}\right)$

$y_{ij}$ ：真实标签。
$\hat{y_{ij}}$ :模型预测概率。

2. 完整案例

导入包

import torch
import torch.nn as nn
import torch.utils.data as Data
import numpy as np
from torch import optim
import random
from tqdm import *
import matplotlib.pyplot as plt

数据集生成

# 数据集生成
soundmark = ['ei',  'bi:',  'si:',  'di:',  'i:',  'ef',  'dʒi:',  'eit∫',  'ai', 'dʒei', 'kei', 'el', 'em', 'en', 'əu', 'pi:', 'kju:',
        'ɑ:', 'es', 'ti:', 'ju:', 'vi:', 'd∧blju:', 'eks', 'wai', 'zi:']
 
alphabet = ['a','b','c','d','e','f','g','h','i','j','k','l','m','n','o','p','q',
         'r','s','t','u','v','w','x','y','z']
 
t = 1000 #总条数
r = 0.9   #扰动项
seq_len = 6
src_tokens, tgt_tokens = [],[] #原始序列、目标序列列表
 
for i in range(t):
    src, tgt = [],[]
    for j in range(seq_len):
        ind = random.randint(0,25)
        src.append(soundmark[ind])
        if random.random() < r:
            tgt.append(alphabet[ind])
        else:
            tgt.append(alphabet[random.randint(0,25)])
    src_tokens.append(src)
    tgt_tokens.append(tgt)
src_tokens[:2], tgt_tokens[:2]

from collections import Counter  # 计数类
 
flatten = lambda l: [item for sublist in l for item in sublist]  # 展平数组
# 构建词表
class Vocab:
    def __init__(self, tokens):
        self.tokens = tokens  # 传入的tokens是二维列表
        self.token2index = {'<pad>': 0, '<bos>': 1, '<eos>': 2, '<unk>': 3}  # 先存好特殊词元
        # 将词元按词频排序后生成列表
        self.token2index.update({
            token: index + 4
            for index, (token, freq) in enumerate(
                sorted(Counter(flatten(self.tokens)).items(), key=lambda x: x[1], reverse=True))
        })
        # 构建id到词元字典
        self.index2token = {index: token for token, index in self.token2index.items()}
 
    def __getitem__(self, query):
        # 单一索引
        if isinstance(query, (str, int)):
            if isinstance(query, str):
                return self.token2index.get(query, 3)
            elif isinstance(query, (int)):
                return self.index2token.get(query, '<unk>')
        # 数组索引
        elif isinstance(query, (list, tuple)):
            return [self.__getitem__(item) for item in query]
 
    def __len__(self):
        return len(self.index2token)

构造数据集

from torch.utils.data import DataLoader, TensorDataset
 
#实例化source和target词表
src_vocab, tgt_vocab = Vocab(src_tokens), Vocab(tgt_tokens)
src_vocab_size = len(src_vocab)  # 源语言词表大小
tgt_vocab_size = len(tgt_vocab)  # 目标语言词表大小
 
#增加开始标识<bos>和结尾标识<eos>
encoder_input = torch.tensor([src_vocab[line + ['<pad>']] for line in src_tokens])
decoder_input = torch.tensor([tgt_vocab[['<bos>'] + line] for line in tgt_tokens])
decoder_output = torch.tensor([tgt_vocab[line + ['<eos>']] for line in tgt_tokens])
 
# 训练集和测试集比例8比2，batch_size = 16
train_size = int(len(encoder_input) * 0.8)
test_size = len(encoder_input) - train_size
batch_size = 16
 
# 自定义数据集函数
class MyDataSet(Data.Dataset):
    def __init__(self, enc_inputs, dec_inputs, dec_outputs):
        super(MyDataSet, self).__init__()
        self.enc_inputs = enc_inputs
        self.dec_inputs = dec_inputs
        self.dec_outputs = dec_outputs
 
    def __len__(self):
        return self.enc_inputs.shape[0]
 
    def __getitem__(self, idx):
        return self.enc_inputs[idx], self.dec_inputs[idx], self.dec_outputs[idx]
 
train_loader = DataLoader(MyDataSet(encoder_input[:train_size], decoder_input[:train_size], decoder_output[:train_size]), batch_size=batch_size)
test_loader = DataLoader(MyDataSet(encoder_input[-test_size:], decoder_input[-test_size:], decoder_output[-test_size:]), batch_size=1)

位置编码

def get_sinusoid_encoding_table(n_position, d_model):
    def cal_angle(position, hid_idx):
        return position / np.power(10000, 2 * (hid_idx // 2) / d_model)
    def get_posi_angle_vec(position):
        return [cal_angle(position, hid_j) for hid_j in range(d_model)]
    sinusoid_table = np.array([get_posi_angle_vec(pos_i) for pos_i in range(n_position)])
    sinusoid_table[:, 0::2] = np.sin(sinusoid_table[:, 0::2]) # 偶数位用正弦函数
    sinusoid_table[:, 1::2] = np.cos(sinusoid_table[:, 1::2]) # 奇数位用余弦函数
    return torch.FloatTensor(sinusoid_table)
print(get_sinusoid_encoding_table(30, 512))

掩码操作

# mask掉没有意义的占位符
def get_attn_pad_mask(seq_q, seq_k):                       # seq_q: [batch_size, seq_len] ,seq_k: [batch_size, seq_len]
    batch_size, len_q = seq_q.size()
    batch_size, len_k = seq_k.size()
    pad_attn_mask = seq_k.data.eq(0).unsqueeze(1)          # 判断 输入那些含有P(=0),用1标记 ,[batch_size, 1, len_k]
    return pad_attn_mask.expand(batch_size, len_q, len_k)
 
# mask掉未来信息
def get_attn_subsequence_mask(seq):                               # seq: [batch_size, tgt_len]
    attn_shape = [seq.size(0), seq.size(1), seq.size(1)]
    subsequence_mask = np.triu(np.ones(attn_shape), k=1)          # 生成上三角矩阵,[batch_size, tgt_len, tgt_len]
    subsequence_mask = torch.from_numpy(subsequence_mask).byte()  #  [batch_size, tgt_len, tgt_len]
    return subsequence_mask

注意力计算函数

# 缩放点积注意力计算
class ScaledDotProductAttention(nn.Module):
    def __init__(self):
        super(ScaledDotProductAttention, self).__init__()
    def forward(self, Q, K, V, attn_mask):
        '''
        Q: [batch_size, n_heads, len_q, d_k]
        K: [batch_size, n_heads, len_k, d_k]
        V: [batch_size, n_heads, len_v(=len_k), d_v]
        attn_mask: [batch_size, n_heads, seq_len, seq_len]
        '''
        scores = torch.matmul(Q, K.transpose(-1, -2)) / np.sqrt(d_k) # scores : [batch_size, n_heads, len_q, len_k]
        scores.masked_fill_(attn_mask, -1e9) # Fills elements of self tensor with value where mask is True.
        attn = nn.Softmax(dim=-1)(scores)
        context = torch.matmul(attn, V) # [batch_size, n_heads, len_q, d_v]
        return context, attn
 
#多头注意力计算
class MultiHeadAttention(nn.Module):
    def __init__(self):
        super(MultiHeadAttention, self).__init__()
        self.W_Q = nn.Linear(d_model, d_k * n_heads, bias=False)
        self.W_K = nn.Linear(d_model, d_k * n_heads, bias=False)
        self.W_V = nn.Linear(d_model, d_v * n_heads, bias=False)
        self.fc = nn.Linear(n_heads * d_v, d_model, bias=False)
    def forward(self, input_Q, input_K, input_V, attn_mask):
        '''
        input_Q: [batch_size, len_q, d_model]
        input_K: [batch_size, len_k, d_model]
        input_V: [batch_size, len_v(=len_k), d_model]
        attn_mask: [batch_size, seq_len, seq_len]
        '''
        residual, batch_size = input_Q, input_Q.size(0)
        # (B, S, D) -proj-> (B, S, D_new) -split-> (B, S, H, W) -trans-> (B, H, S, W)
        Q = self.W_Q(input_Q).view(batch_size, -1, n_heads, d_k).transpose(1,2) # Q: [batch_size, n_heads, len_q, d_k]
        K = self.W_K(input_K).view(batch_size, -1, n_heads, d_k).transpose(1,2) # K: [batch_size, n_heads, len_k, d_k]
        V = self.W_V(input_V).view(batch_size, -1, n_heads, d_v).transpose(1,2) # V: [batch_size, n_heads, len_v(=len_k), d_v]
        attn_mask = attn_mask.unsqueeze(1).repeat(1, n_heads, 1, 1) # attn_mask : [batch_size, n_heads, seq_len, seq_len]
        # context: [batch_size, n_heads, len_q, d_v], attn: [batch_size, n_heads, len_q, len_k]
        context, attn = ScaledDotProductAttention()(Q, K, V, attn_mask)
        context = context.transpose(1, 2).reshape(batch_size, -1, n_heads * d_v) # context: [batch_size, len_q, n_heads * d_v]
        output = self.fc(context) # [batch_size, len_q, d_model]
        return nn.LayerNorm(d_model)(output + residual), attn

前馈神经网络

class PoswiseFeedForwardNet(nn.Module):
    def __init__(self):
        super(PoswiseFeedForwardNet, self).__init__()
        self.fc = nn.Sequential(
            nn.Linear(d_model, d_ff, bias=False),
            nn.ReLU(),
            nn.Linear(d_ff, d_model, bias=False))
        
    def forward(self, inputs):                             # inputs: [batch_size, seq_len, d_model]
        residual = inputs
        output = self.fc(inputs)
        return nn.LayerNorm(d_model)(output + residual)   # 残差 + LayerNorm

编码器与解码器

# 编码器层
class EncoderLayer(nn.Module):
    def __init__(self):
        super(EncoderLayer, self).__init__()
        self.enc_self_attn = MultiHeadAttention()  # 多头注意力
        self.pos_ffn = PoswiseFeedForwardNet()  # 前馈网络
    def forward(self, enc_inputs, enc_self_attn_mask):
        '''
        enc_inputs: [batch_size, src_len, d_model]
        enc_self_attn_mask: [batch_size, src_len, src_len]
        '''
        # enc_outputs: [batch_size, src_len, d_model], attn: [batch_size, n_heads, src_len, src_len]
        enc_outputs, attn = self.enc_self_attn(enc_inputs, enc_inputs, enc_inputs, enc_self_attn_mask) # enc_inputs to same Q,K,V
        enc_outputs = self.pos_ffn(enc_outputs) # enc_outputs: [batch_size, src_len, d_model]
        return enc_outputs, attn
 
# 编码器模块
class Encoder(nn.Module):
    def __init__(self):
        super(Encoder, self).__init__()
        self.src_emb = nn.Embedding(src_vocab_size, d_model)
        self.pos_emb = nn.Embedding.from_pretrained(get_sinusoid_encoding_table(src_vocab_size, d_model), freeze=True)
        self.layers = nn.ModuleList([EncoderLayer() for _ in range(n_layers)])
    def forward(self, enc_inputs):
        '''
        enc_inputs: [batch_size, src_len]
        '''
        word_emb = self.src_emb(enc_inputs) # [batch_size, src_len, d_model]
        pos_emb = self.pos_emb(enc_inputs) # [batch_size, src_len, d_model]
        enc_outputs = word_emb + pos_emb
        enc_self_attn_mask = get_attn_pad_mask(enc_inputs, enc_inputs) # [batch_size, src_len, src_len]
        enc_self_attns = []
        for layer in self.layers:
            # enc_outputs: [batch_size, src_len, d_model], enc_self_attn: [batch_size, n_heads, src_len, src_len]
            enc_outputs, enc_self_attn = layer(enc_outputs, enc_self_attn_mask)
            enc_self_attns.append(enc_self_attn)
        return enc_outputs, enc_self_attns

# 解码器层
class DecoderLayer(nn.Module):
    def __init__(self):
        super(DecoderLayer, self).__init__()
        self.dec_self_attn = MultiHeadAttention()
        self.dec_enc_attn = MultiHeadAttention()
        self.pos_ffn = PoswiseFeedForwardNet()
    def forward(self, dec_inputs, enc_outputs, dec_self_attn_mask, dec_enc_attn_mask):
        '''
        dec_inputs: [batch_size, tgt_len, d_model]
        enc_outputs: [batch_size, src_len, d_model]
        dec_self_attn_mask: [batch_size, tgt_len, tgt_len]
        dec_enc_attn_mask: [batch_size, tgt_len, src_len]
        '''
        # dec_outputs: [batch_size, tgt_len, d_model], dec_self_attn: [batch_size, n_heads, tgt_len, tgt_len]
        dec_outputs, dec_self_attn = self.dec_self_attn(dec_inputs, dec_inputs, dec_inputs, dec_self_attn_mask)
        # dec_outputs: [batch_size, tgt_len, d_model], dec_enc_attn: [batch_size, h_heads, tgt_len, src_len]
        dec_outputs, dec_enc_attn = self.dec_enc_attn(dec_outputs, enc_outputs, enc_outputs, dec_enc_attn_mask)
        dec_outputs = self.pos_ffn(dec_outputs) # [batch_size, tgt_len, d_model]
        return dec_outputs, dec_self_attn, dec_enc_attn
 
# 解码器模块
class Decoder(nn.Module):
    def __init__(self):
        super(Decoder, self).__init__()
        self.tgt_emb = nn.Embedding(tgt_vocab_size, d_model)
        self.pos_emb = nn.Embedding.from_pretrained(get_sinusoid_encoding_table(tgt_vocab_size, d_model),freeze=True)
        self.layers = nn.ModuleList([DecoderLayer() for _ in range(n_layers)])
    def forward(self, dec_inputs, enc_inputs, enc_outputs):
        '''
        dec_inputs: [batch_size, tgt_len]
        enc_intpus: [batch_size, src_len]
        enc_outputs: [batsh_size, src_len, d_model]
        '''
        word_emb = self.tgt_emb(dec_inputs) # [batch_size, tgt_len, d_model]
        pos_emb = self.pos_emb(dec_inputs) # [batch_size, tgt_len, d_model]
        dec_outputs = word_emb + pos_emb
        dec_self_attn_pad_mask = get_attn_pad_mask(dec_inputs, dec_inputs) # [batch_size, tgt_len, tgt_len]
        dec_self_attn_subsequent_mask = get_attn_subsequence_mask(dec_inputs) # [batch_size, tgt_len]
        dec_self_attn_mask = torch.gt((dec_self_attn_pad_mask + dec_self_attn_subsequent_mask), 0) # [batch_size, tgt_len, tgt_len]
        dec_enc_attn_mask = get_attn_pad_mask(dec_inputs, enc_inputs) # [batc_size, tgt_len, src_len]
        dec_self_attns, dec_enc_attns = [], []
        for layer in self.layers:
            # dec_outputs: [batch_size, tgt_len, d_model], dec_self_attn: [batch_size, n_heads, tgt_len, tgt_len], dec_enc_attn: [batch_size, h_heads, tgt_len,src_len]
            dec_outputs, dec_self_attn, dec_enc_attn = layer(dec_outputs, enc_outputs, dec_self_attn_mask, dec_enc_attn_mask)
            dec_self_attns.append(dec_self_attn)
            dec_enc_attns.append(dec_enc_attn)
        return dec_outputs, dec_self_attns, dec_enc_attns

transformer模型

class Transformer(nn.Module):
    def __init__(self):
        super(Transformer, self).__init__()
        self.encoder = Encoder()
        self.decoder = Decoder()
        self.projection = nn.Linear(d_model, tgt_vocab_size, bias=False)
    def forward(self, enc_inputs, dec_inputs):
        '''
        enc_inputs: [batch_size, src_len]
        dec_inputs: [batch_size, tgt_len]
        '''
        # tensor to store decoder outputs
        # outputs = torch.zeros(batch_size, tgt_len, tgt_vocab_size).to(self.device)
        # enc_outputs: [batch_size, src_len, d_model], enc_self_attns: [n_layers, batch_size, n_heads, src_len, src_len]
        enc_outputs, enc_self_attns = self.encoder(enc_inputs)
        # dec_outpus: [batch_size, tgt_len, d_model], dec_self_attns: [n_layers, batch_size, n_heads, tgt_len, tgt_len], dec_enc_attn: [n_layers, batch_size, tgt_len, src_len]
        dec_outputs, dec_self_attns, dec_enc_attns = self.decoder(dec_inputs, enc_inputs, enc_outputs)
        dec_logits = self.projection(dec_outputs) # dec_logits: [batch_size, tgt_len, tgt_vocab_size]
        return dec_logits.view(-1, dec_logits.size(-1)), enc_self_attns, dec_self_attns, dec_enc_attns

模型训练

d_model = 512   # 字 Embedding 的维度
d_ff = 2048     # 前向传播隐藏层维度
d_k = d_v = 64  # K(=Q), V的维度 
n_layers = 6    # 有多少个encoder和decoder
n_heads = 8     # Multi-Head Attention设置为8
num_epochs = 50 # 训练50轮
# 记录损失变化
loss_history = []
 
model = Transformer()
criterion = nn.CrossEntropyLoss(ignore_index=0)
optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.99)
 
for epoch in tqdm(range(num_epochs)):
    total_loss = 0
    for enc_inputs, dec_inputs, dec_outputs in train_loader:
        '''
        enc_inputs: [batch_size, src_len]
        dec_inputs: [batch_size, tgt_len]
        dec_outputs: [batch_size, tgt_len]
        '''
        # enc_inputs, dec_inputs, dec_outputs = enc_inputs.to(device), dec_inputs.to(device), dec_outputs.to(device)
        # outputs: [batch_size * tgt_len, tgt_vocab_size]
        outputs, enc_self_attns, dec_self_attns, dec_enc_attns = model(enc_inputs, dec_inputs)
        loss = criterion(outputs, dec_outputs.view(-1))
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()
        total_loss += loss.item()
    avg_loss = total_loss/len(train_loader)
    loss_history.append(avg_loss)
    print('Epoch:', '%d' % (epoch + 1), 'loss =', '{:.6f}'.format(avg_loss))