.. _sec_sentiment_cnn:
Phân tích tình cảm: Sử dụng mạng thần kinh phức tạp
===================================================
Trong :numref:`chap_cnn`, chúng tôi đã điều tra các cơ chế xử lý dữ
liệu hình ảnh hai chiều với CNN hai chiều, được áp dụng cho các tính
năng cục bộ như pixel liền kề. Mặc dù ban đầu được thiết kế cho tầm nhìn
máy tính, CNN cũng được sử dụng rộng rãi để xử lý ngôn ngữ tự nhiên. Nói
một cách đơn giản, chỉ cần nghĩ về bất kỳ chuỗi văn bản nào như một hình
ảnh một chiều. Bằng cách này, CNN một chiều có thể xử lý các tính năng
địa phương như :math:`n`-gram trong văn bản.
Trong phần này, chúng ta sẽ sử dụng mô hình *textCNN* để chứng minh cách
thiết kế kiến trúc CNN để đại diện cho văn bản duy nhất
:cite:`Kim.2014`. So với :numref:`fig_nlp-map-sa-rnn` sử dụng một
kiến trúc RNN với Glove pretraining để phân tích tình cảm, sự khác biệt
duy nhất trong :numref:`fig_nlp-map-sa-cnn` nằm ở sự lựa chọn của kiến
trúc.
.. _fig_nlp-map-sa-cnn:
.. figure:: ../img/nlp-map-sa-cnn.svg
This section feeds pretrained GloVe to a CNN-based architecture for
sentiment analysis.
.. raw:: html
.. raw:: html
.. code:: python
from mxnet import gluon, init, np, npx
from mxnet.gluon import nn
from d2l import mxnet as d2l
npx.set_np()
batch_size = 64
train_iter, test_iter, vocab = d2l.load_data_imdb(batch_size)
.. parsed-literal::
:class: output
Downloading ../data/aclImdb_v1.tar.gz from http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz...
.. raw:: html
.. raw:: html
.. code:: python
import torch
from torch import nn
from d2l import torch as d2l
batch_size = 64
train_iter, test_iter, vocab = d2l.load_data_imdb(batch_size)
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
def corr1d(X, K):
w = K.shape[0]
Y = np.zeros((X.shape[0] - w + 1))
for i in range(Y.shape[0]):
Y[i] = (X[i: i + w] * K).sum()
return Y
.. raw:: html
.. raw:: html
.. code:: python
def corr1d(X, K):
w = K.shape[0]
Y = torch.zeros((X.shape[0] - w + 1))
for i in range(Y.shape[0]):
Y[i] = (X[i: i + w] * K).sum()
return Y
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
X, K = np.array([0, 1, 2, 3, 4, 5, 6]), np.array([1, 2])
corr1d(X, K)
.. parsed-literal::
:class: output
array([ 2., 5., 8., 11., 14., 17.])
.. raw:: html
.. raw:: html
.. code:: python
X, K = torch.tensor([0, 1, 2, 3, 4, 5, 6]), torch.tensor([1, 2])
corr1d(X, K)
.. parsed-literal::
:class: output
tensor([ 2., 5., 8., 11., 14., 17.])
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
def corr1d_multi_in(X, K):
# First, iterate through the 0th dimension (channel dimension) of `X` and
# `K`. Then, add them together
return sum(corr1d(x, k) for x, k in zip(X, K))
X = np.array([[0, 1, 2, 3, 4, 5, 6],
[1, 2, 3, 4, 5, 6, 7],
[2, 3, 4, 5, 6, 7, 8]])
K = np.array([[1, 2], [3, 4], [-1, -3]])
corr1d_multi_in(X, K)
.. parsed-literal::
:class: output
array([ 2., 8., 14., 20., 26., 32.])
.. raw:: html
.. raw:: html
.. code:: python
def corr1d_multi_in(X, K):
# First, iterate through the 0th dimension (channel dimension) of `X` and
# `K`. Then, add them together
return sum(corr1d(x, k) for x, k in zip(X, K))
X = torch.tensor([[0, 1, 2, 3, 4, 5, 6],
[1, 2, 3, 4, 5, 6, 7],
[2, 3, 4, 5, 6, 7, 8]])
K = torch.tensor([[1, 2], [3, 4], [-1, -3]])
corr1d_multi_in(X, K)
.. parsed-literal::
:class: output
tensor([ 2., 8., 14., 20., 26., 32.])
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
class TextCNN(nn.Block):
def __init__(self, vocab_size, embed_size, kernel_sizes, num_channels,
**kwargs):
super(TextCNN, self).__init__(**kwargs)
self.embedding = nn.Embedding(vocab_size, embed_size)
# The embedding layer not to be trained
self.constant_embedding = nn.Embedding(vocab_size, embed_size)
self.dropout = nn.Dropout(0.5)
self.decoder = nn.Dense(2)
# The max-over-time pooling layer has no parameters, so this instance
# can be shared
self.pool = nn.GlobalMaxPool1D()
# Create multiple one-dimensional convolutional layers
self.convs = nn.Sequential()
for c, k in zip(num_channels, kernel_sizes):
self.convs.add(nn.Conv1D(c, k, activation='relu'))
def forward(self, inputs):
# Concatenate two embedding layer outputs with shape (batch size, no.
# of tokens, token vector dimension) along vectors
embeddings = np.concatenate((
self.embedding(inputs), self.constant_embedding(inputs)), axis=2)
# Per the input format of one-dimensional convolutional layers,
# rearrange the tensor so that the second dimension stores channels
embeddings = embeddings.transpose(0, 2, 1)
# For each one-dimensional convolutional layer, after max-over-time
# pooling, a tensor of shape (batch size, no. of channels, 1) is
# obtained. Remove the last dimension and concatenate along channels
encoding = np.concatenate([
np.squeeze(self.pool(conv(embeddings)), axis=-1)
for conv in self.convs], axis=1)
outputs = self.decoder(self.dropout(encoding))
return outputs
.. raw:: html
.. raw:: html
.. code:: python
class TextCNN(nn.Module):
def __init__(self, vocab_size, embed_size, kernel_sizes, num_channels,
**kwargs):
super(TextCNN, self).__init__(**kwargs)
self.embedding = nn.Embedding(vocab_size, embed_size)
# The embedding layer not to be trained
self.constant_embedding = nn.Embedding(vocab_size, embed_size)
self.dropout = nn.Dropout(0.5)
self.decoder = nn.Linear(sum(num_channels), 2)
# The max-over-time pooling layer has no parameters, so this instance
# can be shared
self.pool = nn.AdaptiveAvgPool1d(1)
self.relu = nn.ReLU()
# Create multiple one-dimensional convolutional layers
self.convs = nn.ModuleList()
for c, k in zip(num_channels, kernel_sizes):
self.convs.append(nn.Conv1d(2 * embed_size, c, k))
def forward(self, inputs):
# Concatenate two embedding layer outputs with shape (batch size, no.
# of tokens, token vector dimension) along vectors
embeddings = torch.cat((
self.embedding(inputs), self.constant_embedding(inputs)), dim=2)
# Per the input format of one-dimensional convolutional layers,
# rearrange the tensor so that the second dimension stores channels
embeddings = embeddings.permute(0, 2, 1)
# For each one-dimensional convolutional layer, after max-over-time
# pooling, a tensor of shape (batch size, no. of channels, 1) is
# obtained. Remove the last dimension and concatenate along channels
encoding = torch.cat([
torch.squeeze(self.relu(self.pool(conv(embeddings))), dim=-1)
for conv in self.convs], dim=1)
outputs = self.decoder(self.dropout(encoding))
return outputs
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
embed_size, kernel_sizes, nums_channels = 100, [3, 4, 5], [100, 100, 100]
devices = d2l.try_all_gpus()
net = TextCNN(len(vocab), embed_size, kernel_sizes, nums_channels)
net.initialize(init.Xavier(), ctx=devices)
.. raw:: html
.. raw:: html
.. code:: python
embed_size, kernel_sizes, nums_channels = 100, [3, 4, 5], [100, 100, 100]
devices = d2l.try_all_gpus()
net = TextCNN(len(vocab), embed_size, kernel_sizes, nums_channels)
def init_weights(m):
if type(m) in (nn.Linear, nn.Conv1d):
nn.init.xavier_uniform_(m.weight)
net.apply(init_weights);
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
glove_embedding = d2l.TokenEmbedding('glove.6b.100d')
embeds = glove_embedding[vocab.idx_to_token]
net.embedding.weight.set_data(embeds)
net.constant_embedding.weight.set_data(embeds)
net.constant_embedding.collect_params().setattr('grad_req', 'null')
.. raw:: html
.. raw:: html
.. code:: python
glove_embedding = d2l.TokenEmbedding('glove.6b.100d')
embeds = glove_embedding[vocab.idx_to_token]
net.embedding.weight.data.copy_(embeds)
net.constant_embedding.weight.data.copy_(embeds)
net.constant_embedding.weight.requires_grad = False
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
lr, num_epochs = 0.001, 5
trainer = gluon.Trainer(net.collect_params(), 'adam', {'learning_rate': lr})
loss = gluon.loss.SoftmaxCrossEntropyLoss()
d2l.train_ch13(net, train_iter, test_iter, loss, trainer, num_epochs, devices)
.. parsed-literal::
:class: output
loss 0.092, train acc 0.969, test acc 0.864
3537.7 examples/sec on [gpu(0), gpu(1)]
.. figure:: output_sentiment-analysis-cnn_900d1d_66_1.svg
.. raw:: html
.. raw:: html
.. code:: python
lr, num_epochs = 0.001, 5
trainer = torch.optim.Adam(net.parameters(), lr=lr)
loss = nn.CrossEntropyLoss(reduction="none")
d2l.train_ch13(net, train_iter, test_iter, loss, trainer, num_epochs, devices)
.. parsed-literal::
:class: output
loss 0.064, train acc 0.979, test acc 0.875
3284.0 examples/sec on [device(type='cuda', index=0), device(type='cuda', index=1)]
.. figure:: output_sentiment-analysis-cnn_900d1d_69_1.svg
.. raw:: html
.. raw:: html
.. raw:: html
.. code:: python
d2l.predict_sentiment(net, vocab, 'this movie is so great')
.. parsed-literal::
:class: output
'positive'
.. code:: python
d2l.predict_sentiment(net, vocab, 'this movie is so bad')
.. parsed-literal::
:class: output
'negative'
.. raw:: html
.. raw:: html
.. code:: python
d2l.predict_sentiment(net, vocab, 'this movie is so great')
.. parsed-literal::
:class: output
'positive'
.. code:: python
d2l.predict_sentiment(net, vocab, 'this movie is so bad')
.. parsed-literal::
:class: output
'negative'
.. raw:: html
.. raw:: html
.. raw:: html
`Discussions `__
.. raw:: html
.. raw:: html
`Discussions `__
.. raw:: html
.. raw:: html