使用Keras进行文本分类。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
准备数据集
出于演示目的,我们将使用 20个新闻组 数据集。数据分为20个类别,我们的工作是预测这些类别。如下所示:文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
通常,对于深度学习,我们将训练和测试数据分开。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
导入所需的软件包
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9 | import pandas as pd文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html import numpy as np文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html import pickle文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html from keras.preprocessing.text import Tokenizer文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html from keras.models import Sequential文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html from keras.layers import Activation, Dense, Dropout文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html from sklearn.preprocessing import LabelBinarizer文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html import sklearn.datasets as skds文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html from pathlib import Path |
将数据从文件加载到Python变量
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 13文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 14文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 15文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 16文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 17文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 18文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 19文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 20文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 21文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 22文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 23 | # For reproducibility文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html np.random.seed(1237)文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html label_index = files_train.target文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html label_names = files_train.target_names文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html labelled_files = files_train.filenames文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html data_tags = ["filename","category","news"]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html data_list = []文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html # Read and add data from file to a list文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html data = pd.DataFrame.from_records(data_list, columns=data_tags) |
在我们的情况下,数据无法以CSV格式提供。我们有文本数据文件,文件存放的目录是我们的标签或类别。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
我们将使用scikit-learn load_files方法。这种方法可以为我们提供原始数据以及标签和标签索引。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
在以上代码的结尾,我们将有一个数据框,其中包含文件名,类别和实际数据。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
拆分数据进行训练和测试
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10 | # lets take 80% data as training and remaining 20% for test.文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html train_size = int(len(data) * .8)文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html train_posts = data['news'][:train_size]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html train_tags = data['category'][:train_size]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html train_files_names = data['filename'][:train_size]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html test_posts = data['news'][train_size:]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html test_tags = data['category'][train_size:]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html test_files_names = data['filename'][train_size:] |
标记化并准备词汇
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 13文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 14文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 15文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 16 | # 20 news groups文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html num_labels = 20文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html vocab_size = 15000文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html batch_size = 100文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html # define Tokenizer with Vocab Size文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html tokenizer = Tokenizer(num_words=vocab_size)文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html tokenizer.fit_on_texts(train_posts) |
在对文本进行分类时,我们首先使用Bag Of Words方法对文本进行预处理。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
预处理输出标签/类
在将文本转换为数字向量后,我们还需要确保标签以神经网络模型接受的数字格式表示。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
建立Keras模型并拟合
PowerShell文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 13文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 14文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 15文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 16文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 17文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 18文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 19文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 20 | model = Sequential() |
它为输入数据的形状以及构成模型的图层类型提供了简单的配置。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
这是一些适合度和测试准确性的代码段文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12 | 100/8145 [..............................] - ETA: 31s - loss: 1.0746e-04 - acc: 1.0000文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 200/8145 [..............................] - ETA: 31s - loss: 0.0186 - acc: 0.9950文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 300/8145 [>.............................] - ETA: 35s - loss: 0.0125 - acc: 0.9967文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 400/8145 [>.............................] - ETA: 32s - loss: 0.0094 - acc: 0.9975文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 500/8145 [>.............................] - ETA: 30s - loss: 0.0153 - acc: 0.9960文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html ...文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7900/8145 [============================>.] - ETA: 0s - loss: 0.1256 - acc: 0.9854文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8000/8145 [============================>.] - ETA: 0s - loss: 0.1261 - acc: 0.9855文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8100/8145 [============================>.] - ETA: 0s - loss: 0.1285 - acc: 0.9854文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8145/8145 [==============================] - 29s 4ms/step - loss: 0.1293 - acc: 0.9854 - val_loss: 1.0597 - val_acc: 0.8742文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html Test accuracy: |
评估模型
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 13 | for i in range(10):文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html prediction = model.predict(np.array([x_test[i]]))文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html predicted_label = text_labels[np.argmax(prediction[0])]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html print(test_files_names.iloc[i])文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html print('Actual label:' + test_tags.iloc[i])文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html print("Predicted label: " + predicted_label) |
在Fit方法训练了我们的数据集之后,我们将如上所述评估模型。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
混淆矩阵
混淆矩阵是可视化模型准确性的最佳方法之一。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
保存模型文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
通常,深度学习的用例就像在不同的会话中进行数据训练,而使用训练后的模型进行预测一样。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6 | # creates a HDF5 file ''文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html model.model.save('')文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html # Save Tokenizer i.e. Vocabulary文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html with open('', 'wb') as handle:文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html pickle.dump(tokenizer, handle, protocol=pickle.HIGHEST_PROTOCOL) |
Keras没有任何实用程序方法可将Tokenizer与模型一起保存。我们必须单独序列化它。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
加载Keras模型
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
预测环境还需要注意标签。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1 | encoder.classes_ #LabelBinarizer |
预测
如前所述,我们已经预留了一些文件进行实际测试。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
Python文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 4文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 5文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 6文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 7文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 8文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 9文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 10文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 11文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 12文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 13文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 14文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 15文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 16文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 17文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 18文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 19文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 20文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 21文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 22文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 23文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 24文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 25文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 26文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 27文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 28 | 文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html labels = np.array(['', '', '',文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html '', '', '',文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html '', '', '', '',文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html '', '', '', '', '',文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html '', '', '',文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html '', ''])文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html ...文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html for x_t in x_tokenized:文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html prediction = model.predict(np.array([x_t]))文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html predicted_label = labels[np.argmax(prediction[0])]文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html print("File ->", test_files[i], "Predicted label: " + predicted_label)文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html i += 1 |
输出量
1文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 2文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html 3 | File -> C:\DL\20news-bydate\20news-bydate-test\\38758 Predicted label:文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html File -> C:\DL\20news-bydate\20news-bydate-test\\76115 Predicted label:文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html File -> C:\DL\20news-bydate\20news-bydate-test\\21329 Predicted label: |
我们知道目录名是文件的真实标签,因此上述预测是准确的。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html
结论
在本文中,我们使用Keras python库构建了一个简单而强大的神经网络。文章源自菜鸟学院-https://www.cainiaoxueyuan.com/ai/17862.html