TensorFlow2.0教程-文字分類

Tensorflow 2.0 教程持續更新： https://blog.csdn.net/qq_31456593/article/details/88606284

TensorFlow 2.0 教程- Keras 快速入門
TensorFlow 2.0 教程-keras 函式api
TensorFlow 2.0 教程-使用keras訓練模型
TensorFlow 2.0 教程-用keras構建自己的網路層
TensorFlow 2.0 教程-keras模型儲存和序列化
TensorFlow 2.0 教程-eager模式
TensorFlow 2.0 教程-Variables
TensorFlow 2.0 教程–AutoGraph

TensorFlow 2.0 深度學習實踐

TensorFlow2.0 教程-影象分類
TensorFlow2.0 教程-文字分類
TensorFlow2.0 教程-過擬合和欠擬合

完整tensorflow2.0教程程式碼請看tensorflow2.0：中文教程tensorflow2_tutorials_chinese(歡迎star)

我們將構建一個簡單的文字分類器，並使用IMDB進行訓練和測試

from __future__ import absolute_import, division, print_function
import tensorflow as tf
from tensorflow import keras

import numpy as np

print(tf.__version__)

2.0.0-alpha0

1.IMDB資料集

下載

imdb=keras.datasets.imdb
(train_x, train_y), (test_x, text_y)=keras.datasets.imdb.load_data(num_words=10000)

瞭解IMDB資料

print("Training entries: {}, labels: {}".format(len(train_x), len(train_y)))

print(train_x[0])
print('len: ',len(train_x[0]), len(train_x[1]))

Training entries: 25000, labels: 25000
[1, 14, 22, 16, 43, 530, 973, 1622, 1385, 65, 458, 4468, 66, 3941, 4, 173, 36, 256, 5, 25, 100, 43, 838, 112, 50, 670, 2, 9, 35, 480, 284, 5, 150, 4, 172, 112, 167, 2, 336, 385, 39, 4, 172, 4536, 1111, 17, 546, 38, 13, 447, 4, 192, 50, 16, 6, 147, 2025, 19, 14, 22, 4, 1920, 4613, 469, 4, 22, 71, 87, 12, 16, 43, 530, 38, 76, 15, 13, 1247, 4, 22, 17, 515, 17, 12, 16, 626, 18, 2, 5, 62, 386, 12, 8, 316, 8, 106, 5, 4, 2223, 5244, 16, 480, 66, 3785, 33, 4, 130, 12, 16, 38, 619, 5, 25, 124, 51, 36, 135, 48, 25, 1415, 33, 6, 22, 12, 215, 28, 77, 52, 5, 14, 407, 16, 82, 2, 8, 4, 107, 117, 5952, 15, 256, 4, 2, 7, 3766, 5, 723, 36, 71, 43, 530, 476, 26, 400, 317, 46, 7, 4, 2, 1029, 13, 104, 88, 4, 381, 15, 297, 98, 32, 2071, 56, 26, 141, 6, 194, 7486, 18, 4, 226, 22, 21, 134, 476, 26, 480, 5, 144, 30, 5535, 18, 51, 36, 28, 224, 92, 25, 104, 4, 226, 65, 16, 38, 1334, 88, 12, 16, 283, 5, 16, 4472, 113, 103, 32, 15, 16, 5345, 19, 178, 32]
len:  218 189

建立id和詞的匹配字典

word_index = imdb.get_word_index()

word2id = {k:(v+3) for k, v in word_index.items()}
word2id['<PAD>'] = 0
word2id['<START>'] = 1
word2id['<UNK>'] = 2
word2id['<UNUSED>'] = 3

id2word = {v:k for k, v in word2id.items()}
def get_words(sent_ids):
    return ' '.join([id2word.get(i, '?') for i in sent_ids])

sent = get_words(train_x[0])
print(sent)

<START> this film was just brilliant casting location scenery story direction everyone's really suited the part they played and you could just imagine being there robert <UNK> is an amazing actor and now the same being director <UNK> father came from the same scottish island as myself so i loved the fact there was a real connection with this film the witty remarks throughout the film were great it was just brilliant so much that i bought the film as soon as it was released for <UNK> and would recommend it to everyone to watch and the fly fishing was amazing really cried at the end it was so sad and you know what they say if you cry at a film it must have been good and this definitely was also <UNK> to the two little boy's that played the <UNK> of norman and paul they were just brilliant children are often left out of the <UNK> list i think because the stars that play them all grown up are such a big profile for the whole film but these children are amazing and should be praised for what they have done don't you think the whole story was so lovely because it was true and was someone's life after all that was shared with us all

2.準備資料

# 句子末尾padding
train_x = keras.preprocessing.sequence.pad_sequences(
    train_x, value=word2id['<PAD>'],
    padding='post', maxlen=256
)
test_x = keras.preprocessing.sequence.pad_sequences(
    test_x, value=word2id['<PAD>'],
    padding='post', maxlen=256
)
print(train_x[0])
print('len: ',len(train_x[0]), len(train_x[1]))

[   1   14   22   16   43  530  973 1622 1385   65  458 4468   66 3941
    4  173   36  256    5   25  100   43  838  112   50  670    2    9
   35  480  284    5  150    4  172  112  167    2  336  385   39    4
  172 4536 1111   17  546   38   13  447    4  192   50   16    6  147
 2025   19   14   22    4 1920 4613  469    4   22   71   87   12   16
   43  530   38   76   15   13 1247    4   22   17  515   17   12   16
  626   18    2    5   62  386   12    8  316    8  106    5    4 2223
 5244   16  480   66 3785   33    4  130   12   16   38  619    5   25
  124   51   36  135   48   25 1415   33    6   22   12  215   28   77
   52    5   14  407   16   82    2    8    4  107  117 5952   15  256
    4    2    7 3766    5  723   36   71   43  530  476   26  400  317
   46    7    4    2 1029   13  104   88    4  381   15  297   98   32
 2071   56   26  141    6  194 7486   18    4  226   22   21  134  476
   26  480    5  144   30 5535   18   51   36   28  224   92   25  104
    4  226   65   16   38 1334   88   12   16  283    5   16 4472  113
  103   32   15   16 5345   19  178   32    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0    0    0    0    0
    0    0    0    0    0    0    0    0    0    0    0    0    0    0
    0    0    0    0]
len:  256 256

3.構建模型

import tensorflow.keras.layers as layers
vocab_size = 10000
model = keras.Sequential()
model.add(layers.Embedding(vocab_size, 16))
model.add(layers.GlobalAveragePooling1D())
model.add(layers.Dense(16, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))
model.summary()
model.compile(optimizer='adam',
             loss='binary_crossentropy',
             metrics=['accuracy'])

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
embedding (Embedding)        (None, None, 16)          160000    
_________________________________________________________________
global_average_pooling1d (Gl (None, 16)                0         
_________________________________________________________________
dense (Dense)                (None, 16)                272       
_________________________________________________________________
dense_1 (Dense)              (None, 1)                 17        
=================================================================
Total params: 160,289
Trainable params: 160,289
Non-trainable params: 0
_________________________________________________________________

4.模型訓練與驗證

x_val = train_x[:10000]
x_train = train_x[10000:]

y_val = train_y[:10000]
y_train = train_y[10000:]

history = model.fit(x_train,y_train,
                   epochs=40, batch_size=512,
                   validation_data=(x_val, y_val),
                   verbose=1)

result = model.evaluate(test_x, text_y)
print(result)

Train on 15000 samples, validate on 10000 samples
Epoch 1/40
15000/15000 [==============================] - 1s 73us/sample - loss: 0.6919 - accuracy: 0.5071 - val_loss: 0.6901 - val_accuracy: 0.5101
Epoch 2/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.6864 - accuracy: 0.6242 - val_loss: 0.6829 - val_accuracy: 0.6380
Epoch 3/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.6752 - accuracy: 0.6881 - val_loss: 0.6691 - val_accuracy: 0.7091
Epoch 4/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.6559 - accuracy: 0.7162 - val_loss: 0.6471 - val_accuracy: 0.7509
Epoch 5/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.6274 - accuracy: 0.7697 - val_loss: 0.6175 - val_accuracy: 0.7724
Epoch 6/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.5909 - accuracy: 0.8049 - val_loss: 0.5821 - val_accuracy: 0.7869
Epoch 7/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.5490 - accuracy: 0.8208 - val_loss: 0.5418 - val_accuracy: 0.8158
Epoch 8/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.5054 - accuracy: 0.8437 - val_loss: 0.5030 - val_accuracy: 0.8285
Epoch 9/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.4630 - accuracy: 0.8557 - val_loss: 0.4662 - val_accuracy: 0.8400
Epoch 10/40
15000/15000 [==============================] - 1s 49us/sample - loss: 0.4239 - accuracy: 0.8707 - val_loss: 0.4345 - val_accuracy: 0.8470
Epoch 11/40
15000/15000 [==============================] - 1s 46us/sample - loss: 0.3896 - accuracy: 0.8772 - val_loss: 0.4070 - val_accuracy: 0.8563
Epoch 12/40
15000/15000 [==============================] - 1s 47us/sample - loss: 0.3599 - accuracy: 0.8867 - val_loss: 0.3856 - val_accuracy: 0.8594
Epoch 13/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.3352 - accuracy: 0.8925 - val_loss: 0.3660 - val_accuracy: 0.8646
Epoch 14/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.3131 - accuracy: 0.8978 - val_loss: 0.3517 - val_accuracy: 0.8697
Epoch 15/40
15000/15000 [==============================] - 1s 48us/sample - loss: 0.2947 - accuracy: 0.9013 - val_loss: 0.3392 - val_accuracy: 0.8716
Epoch 16/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.2782 - accuracy: 0.9077 - val_loss: 0.3293 - val_accuracy: 0.8747
Epoch 17/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.2632 - accuracy: 0.9126 - val_loss: 0.3208 - val_accuracy: 0.8757
Epoch 18/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.2500 - accuracy: 0.9159 - val_loss: 0.3132 - val_accuracy: 0.8800
Epoch 19/40
15000/15000 [==============================] - 1s 46us/sample - loss: 0.2381 - accuracy: 0.9197 - val_loss: 0.3073 - val_accuracy: 0.8792
Epoch 20/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.2274 - accuracy: 0.9229 - val_loss: 0.3029 - val_accuracy: 0.8801
Epoch 21/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.2167 - accuracy: 0.9277 - val_loss: 0.2992 - val_accuracy: 0.8811
Epoch 22/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.2077 - accuracy: 0.9299 - val_loss: 0.2951 - val_accuracy: 0.8835
Epoch 23/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.1986 - accuracy: 0.9335 - val_loss: 0.2931 - val_accuracy: 0.8827
Epoch 24/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.1907 - accuracy: 0.9371 - val_loss: 0.2911 - val_accuracy: 0.8835
Epoch 25/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.1828 - accuracy: 0.9415 - val_loss: 0.2885 - val_accuracy: 0.8841
Epoch 26/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.1756 - accuracy: 0.9436 - val_loss: 0.2884 - val_accuracy: 0.8840
Epoch 27/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.1689 - accuracy: 0.9463 - val_loss: 0.2870 - val_accuracy: 0.8836
Epoch 28/40
15000/15000 [==============================] - 1s 41us/sample - loss: 0.1624 - accuracy: 0.9497 - val_loss: 0.2870 - val_accuracy: 0.8853
Epoch 29/40
15000/15000 [==============================] - 1s 46us/sample - loss: 0.1568 - accuracy: 0.9523 - val_loss: 0.2872 - val_accuracy: 0.8840
Epoch 30/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.1509 - accuracy: 0.9534 - val_loss: 0.2864 - val_accuracy: 0.8858
Epoch 31/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.1449 - accuracy: 0.9567 - val_loss: 0.2866 - val_accuracy: 0.8858
Epoch 32/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.1395 - accuracy: 0.9595 - val_loss: 0.2874 - val_accuracy: 0.8856
Epoch 33/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.1343 - accuracy: 0.9600 - val_loss: 0.2888 - val_accuracy: 0.8863
Epoch 34/40
15000/15000 [==============================] - 1s 44us/sample - loss: 0.1297 - accuracy: 0.9623 - val_loss: 0.2903 - val_accuracy: 0.8843
Epoch 35/40
15000/15000 [==============================] - 1s 43us/sample - loss: 0.1255 - accuracy: 0.9630 - val_loss: 0.2915 - val_accuracy: 0.8870
Epoch 36/40
15000/15000 [==============================] - 1s 42us/sample - loss: 0.1208 - accuracy: 0.9659 - val_loss: 0.2928 - val_accuracy: 0.8862
Epoch 37/40
15000/15000 [==============================] - 1s 48us/sample - loss: 0.1162 - accuracy: 0.9679 - val_loss: 0.2949 - val_accuracy: 0.8851
Epoch 38/40
15000/15000 [==============================] - 1s 49us/sample - loss: 0.1121 - accuracy: 0.9691 - val_loss: 0.2975 - val_accuracy: 0.8848
Epoch 39/40
15000/15000 [==============================] - 1s 49us/sample - loss: 0.1088 - accuracy: 0.9697 - val_loss: 0.3003 - val_accuracy: 0.8840
Epoch 40/40
15000/15000 [==============================] - 1s 45us/sample - loss: 0.1046 - accuracy: 0.9721 - val_loss: 0.3022 - val_accuracy: 0.8843
25000/25000 [==============================] - 1s 22us/sample - loss: 0.3216 - accuracy: 0.8729
[0.32155542838573453, 0.87292]

5.檢視準確率時序圖

import matplotlib.pyplot as plt
history_dict = history.history
history_dict.keys()
acc = history_dict['accuracy']
val_acc = history_dict['val_accuracy']
loss = history_dict['loss']
val_loss = history_dict['val_loss']
epochs = range(1, len(acc)+1)

plt.plot(epochs, loss, 'bo', label='train loss')
plt.plot(epochs, val_loss, 'b', label='val loss')
plt.title('Train and val loss')
plt.xlabel('Epochs')
plt.xlabel('loss')
plt.legend()
plt.show()

plt.clf()   # clear figure

plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()

plt.show()