第三章 神经网络入门

""" 
路透社数据集(新闻分类)--多分类问题 
"""
# 1.加载路透社数据集
from keras.datasets import reuters
(train_data, train_labels), (test_data, test_labels) = reuters.load_data(num_words=10000)
print(len(train_data))
print(len(test_data))
print(train_data[10])


# 2.将索引解码为新闻文本(预处理)
word_index = reuters.get_word_index()       # 是一个将单词映射为索引的子典
reverse_word_index = dict([(value, key) for (key, value) in word_index.items()],)   # 键值颠倒
decoded_newswisre = ''.join([reverse_word_index.get(i - 3, '?') for i in train_data[0]])
print(train_labels[10])


# 3.编码数据(预处理)
import numpy as np
def vectorize_sequences(sequences, dimension=10000):
    results = np.zeros((len(sequences), dimension))   # 创建一个2D的0矩阵
    for i, sequence in enumerate(sequences):
        results[i, sequence] = 1.                     # 将results[i]的索引设为1
    return results

x_train = vectorize_sequences(train_data)       # 将训练数据向量化
x_test = vectorize_sequences(test_data)         # 将测试数据向量化


def to_one_hot(labels, dimension=46):           # ******** 标签向量化方式1：one_hot编码（分类编码）
    results = np.zeros((len(labels), dimension))
    for i, label in enumerate(labels):
        results[i, label] = 1.
    return results
one_hot_train_labels = to_one_hot(train_labels)     # 将训练标签向量化
one_hot_test_labels = to_one_hot(test_labels)       # 将测试标签向量化


# ******** 标签向量化方式2：Keras内置方法
from keras.utils.np_utils import to_categorical

# one_hot_train_labels = to_categorical(train_labels)
# one_hot_test_labels = to_categorical(test_labels)


# ******** 标签向量化方式3：将标签列表转换为整张张量(用标签编码时，loss='sparse_categorical_crossentropy')
# one_hot_train_labels = np.array(train_labels)
# one_hot_test_labels = np.array(test_labels)



# 4.构建网络
from keras import models
from keras import layers

# 定义模型
model = models.Sequential()
model.add(layers.Dense(64, activation='relu', input_shape=(10000, )))
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(46, activation='softmax'))     # 46分类
model.summary()


# 编译模型
model.compile(optimizer='rmsprop',                  # rmsprop优化器
             loss='categorical_crossentropy',       # 损失函数：分类交叉熵
             # loss='sparse_categorical_crossentropy',       # 用标签编码时
             metrics=['acc'])                       # 指标：精度

# 留出验证集
x_val = x_train[:1000]                          # 验证集数据
partial_x_train = x_train[1000:]                # 训练集数据
y_val = one_hot_train_labels[:1000]             # 验证集标签
partial_y_train = one_hot_train_labels[1000:]   # 训练集标签

# 训练模型
history = model.fit(partial_x_train,                  # 训练数据
                    partial_y_train,
                    epochs=20,                        #  迭代次数20
                    batch_size=512,                   # 512个样本的小批量
                    validation_data=(x_val, y_val))   # 将前1000个验证数据传入validation_data参数来完成


history_dict = history.history  # 是一个字典，包含训练过程中的所有数据
print(history_dict.keys())        # dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])


# 5.绘制训练损失和验证损失
import matplotlib.pyplot as plt

loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(1, len(loss) + 1)

plt.plot(epochs, loss, 'r', label='Training loss')
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('Training and Validation loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()

# 绘制训练精度和验证精度
plt.clf()        # 清空图像

acc = history.history['acc']
val_acc = history.history['val_acc']

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


# 6.从头开始重新训练一个模型
model = models.Sequential()
model.add(layers.Dense(64, activation='relu', input_shape=(10000, )))
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(46, activation='softmax'))         # 46分类

model.compile(optimizer='rmsprop',
              loss='categorical_crossentropy',
              # loss='sparse_categorical_crossentropy',       # 用标签编码时
              metrics=['accuracy'])

model.fit(partial_x_train,
          partial_y_train,
          epochs=9,
          batch_size=512,
          validation_data=(x_val, y_val))        # 将前1000个验证数据传入validation_data参数来完成(验证集))
results = model.evaluate(x_test, one_hot_test_labels)   # 评估测试数据(测试集)
print(results)       # dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])


predictions = model.predict(x_test)  # 在新数据上生成预测结果
print(predictions[0].shape)          # (46,)
print(np.sum(predictions[0]))
print(np.argmax(predictions[0]))


# 基准的方法
import copy
test_labels_copy = copy.copy(test_labels)
np.random.shuffle(test_labels_copy)
hits_array = np.array(test_labels) == np.array(test_labels_copy)
acc = float(np.sum(hits_array)) / len(test_labels)
print('基准方法的精度acc是:', acc)

代码运行结果：