tensorflow 在逻辑回归的例子

from tensorflow.keras.datasets import fashion_mnist
#初始化参数
batch_size = 128
epochs = 20
n_classes = 10
learning_rate = 0.1
width = 28 # of our imagesheight = 28 # of our images
height = 28
x_train = x_train.astype('float32') / 255
x_test = x_test.astype('float32') / 255.
(x_train, y_train),(x_test, y_test)=fashion_mnist.load_data() #加载数据
x_train = x_train.reshape((60000, width * height)) #flatten data
x_test = x_test.reshape((10000, width * height))
split = 50000 
(x_train, x_valid) = x_train[:split], x_train[split:] 
(y_train, y_valid) = y_train[:split], y_train[split:]
y_train_ohe = tf.one_hot(y_train, depth=n_classes).numpy()#one hot 编码
y_valid_ohe = tf.one_hot(y_valid, depth=n_classes).numpy()
y_test_ohe = tf.one_hot(y_test, depth=n_classes).numpy()

#定义模型。
class LogisticRegression(tf.keras.Model):
    def __init__(self, num_classes):
        super().__init__()
        self.dense=tf.keras.layers.Dense(10)
    def call(self,inputs,training=None, mask=None):
        output = self.dense(inputs) 
        with tf.device('/cpu:0'):
            output = tf.nn.softmax(output)
        return output
#compile 模型。
model = LogisticRegression(n_classes)
optimiser =tf.keras.optimizers.Adam()
model.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['accuracy'], )

#
dummy_x = tf.zeros((1, width * height))
model.call(dummy_x)
#保存训练好的模型参数。
from tensorflow.keras.callbacks import ModelCheckpoint
checkpointer=ModelCheckpoint(filepath="./model.weights.best.hdf5", verbose=2, save_best_only=True, save_weights_only=True)
model.fit(x_train, y_train_ohe, batch_size=batch_size, epochs=epochs,validation_data=(x_valid, y_valid_ohe), callbacks=[checkpointer], verbose=2)
#加载最好的模型参数。
model.load_weights("./model.weights.best.hdf5")
#评估模型参数。
scores = model.evaluate(x_test, y_test_ohe, batch_size, verbose=2)
#还原预测的结果。
index_predicted = np.argmax(y_predictions[index]) 

model2 = LogisticRegression(n_classes) #实例化模型对象