Keras 数据集对于训练集和测试集具有不同的向量长度

Question

我正在尝试使用来自 keras.datasets 的路透社和 imdb 数据集。标准调用是：

(x_train, y_train), (x_test, y_test) = imdb.load_data(path="imdb.npz",
                                                  num_words=None,
                                                  skip_top=0,
                                                  maxlen=None,
                                                  seed=113,
                                                  start_char=1,
                                                  oov_char=2,
                                                  index_from=3)

当我查看维度时，火车数据集给出 (25000, 10922)，这很有意义。但是测试给出了（25000，）。如果你转储单个测试数据集元素，如 x_test[0]，它会给出一个列表而不是 numpy.array。问题是每行的列表维度都会发生变化，并且总是与训练向量维度不同。您应该如何将其用作测试数据？

Answer 1

嗯，正如您提到的，x_train 和 x_test 中的每个元素都是一个列表。该列表包含句子（或段落或在本例中为评论）的单词索引，并且由于句子可能具有不同数量的单词，因此相应的表示也具有可变长度。让我们解码其中一个句子，看看它长什么样，并更加熟悉数据集：

from keras.datasets import imdb

(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=10000)

# a mapping from words to their indices, for example `human`: 403
word_index = imdb.get_word_index()

# create the reverse mapping i.e. from indices to words
rev_word_index = {idx:w for w,idx in word_index.items()}

def decode_sentence(s):
    # index 0 to 2 are reserved for things like padding, unknown word, etc.
    decoded_sent = [rev_word_index.get(idx-3, '[RES]') for idx in s]
    return ' '.join(decoded_sent)

print(decode_sentence(x_train[100]))

以人类可读的形式输出所选评论：

[RES] i am a great fan of david lynch and have everything that he's made on dvd except for hotel room the 2 hour twin peaks movie so when i found out about this i immediately grabbed it and and what is this it's a bunch of [RES] drawn black and white cartoons that are loud and foul mouthed and unfunny maybe i don't know what's good but maybe this is just a bunch of crap that was [RES] on the public under the name of david lynch to make a few bucks too let me make it clear that i didn't care about the foul language part but had to keep [RES] the sound because my neighbors might have all in all this is a highly disappointing release and may well have just been left in the [RES] box set as a curiosity i highly recommend you don't spend your money on this 2 out of 10

此数据的使用完全取决于您和您要解决的问题。您可以将句子原样提供给可以处理可变长度句子的网络。这通常由一维卷积层或 LSTM 层或两者的混合组成。另一种方法是通过将所有句子编码为固定长度编码来使所有句子具有相同的长度。这是一个示例，one-hot 将每个句子编码为一个 0 和 1 的向量，所有句子的长度都是固定的：

from keras.datasets import imdb
import numpy as np

# you can limit the vocabulary size by passing `num_words` argument 
# to ignore rare words and make data more manageable
(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=10000)

def encode_sentences(sentences, dim):
    encodings = np.zeros((len(sentences), dim))
    for idx, sentence in enumerate(sentences):
        encodings[idx, sentence] = 1.0
    return encodings

x_train = encode_sentences(x_train, 10000)
x_test = encode_sentences(x_test, 10000)

print(x_train.shape)
print(x_test.shape)

输出：

(25000, 10000)
(25000, 10000)

所有句子都被编码为长度为 10000 的向量，其中该向量的第 i 个元素表示索引为 i 的单词是否出现在相应的句子中。

另一种方法是截断或填充句子以使其长度相同。这是一个例子：

from keras.datasets import imdb
from keras import preprocessing

n_feats = 10000   # maximum number of words we want to consider in our vocabulary
max_len = 500     # maximum length of each sentence (i.e. truncate those longer
                  # than 500 words and pad those shorter than 500 words)

# you can limit the vocabulary size by passing `num_words` argument
# to ignore rare words and make it more manageable
(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=n_feats)

# preprocess the sequences (i.e. truncate or pad)
x_train = preprocessing.sequence.pad_sequences(x_train, maxlen=max_len)
x_test = preprocessing.sequence.pad_sequences(x_test, maxlen=max_len)

print(x_train.shape)
print(x_test.shape)

输出：

(25000, 500)
(25000, 500)

现在25000个句子的长度都一样了，可以用了

我强烈推荐阅读 the Keras documentation on this dataset。