Python数据科学 速查表 模型架构 审视模型

Keras
呆鸟 译 序贯模型 >>>
>>>
model.output_shape
model.summary()
模型输出形状
模型摘要展示
>>> from keras.models import Sequential >>> model.get_config() 模型配置
列出模型的所有权重张量
天善智能 商业智能与大数据社区 www.hellobi.com
>>> model = Sequential() >>> model.get_weights()
>>> model2 = Sequential()
>>> model3 = Sequential() 编译模型

Keras 多层感知器（MLP） 多层感知器：二进制分类

二进制分类 >>> model.compile(optimizer='adam',
loss='binary_crossentropy',
Keras是强大、易用的深度学习库，基于Theano和TensorFlow提供
>>> from keras.layers import Dense metrics=['accuracy'])
>>> model.add(Dense(12, 多层感知器：多级分类
了高阶神经网络API，用于开发和评估深度学习模型。 input_dim=8, >>> model.compile(optimizer='rmsprop',
kernel_initializer='uniform', loss='categorical_crossentropy',

示例 多层感知器：回归
activation='relu')) metrics=['accuracy'])
>>> model.add(Dense(8,kernel_initializer='uniform',activation='relu'))
>>> model.add(Dense(1,kernel_initializer='uniform',activation='sigmoid')) >>> model.compile(optimizer='rmsprop',
>>> import numpy as np loss='mse',
>>> from keras.models import Sequential 多级分类 metrics=['mae'])
>>> from keras.layers import Dense
递归神经网络
>>> from keras.layers import Dropout
>>> data = np.random.random((1000,100)) >>> model.add(Dense(512,activation='relu',input_shape=(784,)))
>>> labels = np.random.randint(2,size=(1000,1)) >>> model.add(Dropout(0.2)) >>> model3.compile(loss='binary_crossentropy',
>>> model = Sequential() optimizer='adam',
>>> model.add(Dense(512,activation='relu')) metrics=['accuracy'])
>>> model.add(Dense(32, >>> model.add(Dropout(0.2))
activation='relu', >>> model.add(Dense(10,activation='softmax'))

>>>
input_dim=100))
model.add(Dense(1, activation='sigmoid')) 回归 模型训练
>>> model.compile(optimizer='rmsprop', >>> model.add(Dense(64,activation='relu',input_dim=train_data.shape[1])) >>> model3.fit(x_train4,
loss='binary_crossentropy', >>> model.add(Dense(1)) y_train4,

卷积神经网络（CNN）
metrics=['accuracy']) batch_size=32,
>>> model.fit(data,labels,epochs=10,batch_size=32) epochs=15,
verbose=1,
>>> predictions = model.predict(data) >>> from keras.layers import Activation,Conv2D,MaxPooling2D,Flatten validation_data=(x_test4,y_test4))
>>> model2.add(Conv2D(32,(3,3),padding='same',input_shape=x_train.shape[1:]))
数据 参阅 NumPy, Pandas & Scikit-Learn >>> model2.add(Activation('relu'))
评估模型性能
>>> model2.add(Conv2D(32,(3,3)))
数据要存为 NumPy 数组或数组列表，使用 sklearn.cross_validation >>> model2.add(Activation('relu')) >>> score = model3.evaluate(x_test,
的 train_test_split 模块进行分割将数据分割为训练集与测试集。
>>> model2.add(MaxPooling2D(pool_size=(2,2))) y_test,
>>> model2.add(Dropout(0.25)) batch_size=32)
>>> model2.add(Conv2D(64,(3,3), padding='same'))
Keras 数据集 >>> model2.add(Activation('relu')) 预测
>>> model2.add(Conv2D(64,(3, 3)))
>>> from keras.datasets import boston_housing, >>> model2.add(Activation('relu')) >>> model3.predict(x_test4, batch_size=32)
mnist, >>> model2.add(MaxPooling2D(pool_size=(2,2))) >>> model3.predict_classes(x_test4,batch_size=32)
cifar10, >>> model2.add(Dropout(0.25))
imdb
>>> (x_train,y_train),(x_test,y_test) = mnist.load_data() >>> model2.add(Flatten()) 保存/加载模型
>>> (x_train2,y_train2),(x_test2,y_test2) = boston_housing.load_data() >>> model2.add(Dense(512))
>>> (x_train3,y_train3),(x_test3,y_test3) = cifar10.load_data() >>> model2.add(Activation('relu')) >>> from keras.models import load_model
>>> (x_train4,y_train4),(x_test4,y_test4) = imdb.load_data(num_words=20000) >>> model2.add(Dropout(0.5)) >>> model3.save('model_file.h5')
>>> num_classes = 10 >>> my_model = load_model('my_model.h5')
>>> model2.add(Dense(num_classes))
其它 >>> model2.add(Activation('softmax'))
模型微调
递归神经网络（RNN）
参数优化
>>> from urllib.request import urlopen
>>> data = np.loadtxt(urlopen("http://archive.ics.uci.edu/ >>> from keras.klayers import Embedding,LSTM
ml/machine-learning-databases/pima-indians-diabetes/
pima-indians-diabetes.data"),delimiter=",") >>> model3.add(Embedding(20000,128)) >>> from keras.optimizers import RMSprop
>>> X = data[:,0:8] >>> model3.add(LSTM(128,dropout=0.2,recurrent_dropout=0.2)) >>> opt = RMSprop(lr=0.0001, decay=1e-6)
>>> y = data [:,8] >>> model3.add(Dense(1,activation='sigmoid')) >>> model2.compile(loss='categorical_crossentropy',
optimizer=opt,
预处理 参阅 NumPy 与 Scikit-Learn
metrics=['accuracy'])

早停法
序列填充 训练与测试集
>>> from keras.callbacks import EarlyStopping
>>> from keras.preprocessing import sequence >>> from sklearn.model_selection import train_test_split >>> early_stopping_monitor = EarlyStopping(patience=2)
>>> x_train4 = sequence.pad_sequences(x_train4,maxlen=80) >>> X_train5,X_test5,y_train5,y_test5 = train_test_split(X, >>> model3.fit(x_train4,
>>> x_test4 = sequence.pad_sequences(x_test4,maxlen=80) y,
test_size=0.33, y_train4,
独热编码
random_state=42) batch_size=32,

标准化/归一化
epochs=15,
>>> from keras.utils import to_categorical validation_data=(x_test4,y_test4),
>>> Y_train = to_categorical(y_train, num_classes) >>> from sklearn.preprocessing import StandardScaler callbacks=[early_stopping_monitor])
>>> Y_test = to_categorical(y_test, num_classes) >>> scaler = StandardScaler().fit(x_train2)
原文作者
>>> Y_train3 = to_categorical(y_train3, num_classes) >>> standardized_X = scaler.transform(x_train2) DataCamp
>>> Y_test3 = to_categorical(y_test3, num_classes) >>> standardized_X_test = scaler.transform(x_test2) Learn Python for Data Science Interactively