在tensorflow的环境下，搭建resnet18，加入SENet进行模型增强，强化数据，训练cifa10,使准确率在90左右，给出完整代码，打印出最终结果

以下是在tensorflow环境下搭建resnet18，并加入SENet进行模型增强的代码：

import tensorflow as tf
from tensorflow import keras

def resnet_block(inputs, filters, kernel_size, strides):
    x = keras.layers.Conv2D(filters, kernel_size=kernel_size, strides=strides, padding='same')(inputs)
    x = keras.layers.BatchNormalization()(x)
    x = keras.layers.ReLU()(x)
    x = keras.layers.Conv2D(filters, kernel_size=kernel_size, strides=1, padding='same')(x)
    x = keras.layers.BatchNormalization()(x)
    if strides == 2:
        inputs = keras.layers.Conv2D(filters, kernel_size=1, strides=strides, padding='same')(inputs)
        inputs = keras.layers.BatchNormalization()(inputs)
    x = keras.layers.add([x, inputs])
    x = keras.layers.ReLU()(x)
    return x

def resnet18(input_shape, num_classes):
    inputs = keras.layers.Input(shape=input_shape)
    x = keras.layers.Conv2D(64, kernel_size=3, strides=1, padding='same')(inputs)
    x = keras.layers.BatchNormalization()(x)
    x = keras.layers.ReLU()(x)
    x = resnet_block(x, filters=64, kernel_size=3, strides=1)
    x = resnet_block(x, filters=64, kernel_size=3, strides=1)
    x = resnet_block(x, filters=128, kernel_size=3, strides=2)
    x = resnet_block(x, filters=128, kernel_size=3, strides=1)
    x = resnet_block(x, filters=256, kernel_size=3, strides=2)
    x = resnet_block(x, filters=256, kernel_size=3, strides=1)
    x = resnet_block(x, filters=512, kernel_size=3, strides=2)
    x = resnet_block(x, filters=512, kernel_size=3, strides=1)
    x = keras.layers.GlobalAveragePooling2D()(x)
    x = keras.layers.Dense(num_classes, activation='softmax')(x)
    model = keras.models.Model(inputs=inputs, outputs=x)
    return model

class SEBlock(keras.layers.Layer):
    def __init__(self, reduction_ratio=16, **kwargs):
        super().__init__(**kwargs)
        self.reduction_ratio = reduction_ratio

    def build(self, input_shape):
        _, _, _, filters = input_shape
        self.average_pooling = keras.layers.GlobalAveragePooling2D()
        self.fc1 = keras.layers.Dense(filters // self.reduction_ratio, activation='relu')
        self.fc2 = keras.layers.Dense(filters, activation='sigmoid')

    def call(self, inputs):
        x = self.average_pooling(inputs)
        x = self.fc1(x)
        x = self.fc2(x)
        x = keras.layers.Reshape((1, 1, filters))(x)
        return inputs * x

def se_resnet18(input_shape, num_classes):
    inputs = keras.layers.Input(shape=input_shape)
    x = keras.layers.Conv2D(64, kernel_size=3, strides=1, padding='same')(inputs)
    x = keras.layers.BatchNormalization()(x)
    x = keras.layers.ReLU()(x)
    x = SEBlock()(x)
    x = resnet_block(x, filters=64, kernel_size=3, strides=1)
    x = SEBlock()(x)
    x = resnet_block(x, filters=64, kernel_size=3, strides=1)
    x = SEBlock()(x)
    x = resnet_block(x, filters=128, kernel_size=3, strides=2)
    x = SEBlock()(x)
    x = resnet_block(x, filters=128, kernel_size=3, strides=1)
    x = SEBlock()(x)
    x = resnet_block(x, filters=256, kernel_size=3, strides=2)
    x = SEBlock()(x)
    x = resnet_block(x, filters=256, kernel_size=3, strides=1)
    x = SEBlock()(x)
    x = resnet_block(x, filters=512, kernel_size=3, strides=2)
    x = SEBlock()(x)
    x = resnet_block(x, filters=512, kernel_size=3, strides=1)
    x = SEBlock()(x)
    x = keras.layers.GlobalAveragePooling2D()(x)
    x = keras.layers.Dense(num_classes, activation='softmax')(x)
    model = keras.models.Model(inputs=inputs, outputs=x)
    return model

接下来，我们可以使用cifa10数据集进行训练，并打印出最终结果：

import numpy as np
from tensorflow.keras.datasets import cifar10
from tensorflow.keras.preprocessing.image import ImageDataGenerator

# 加载数据集
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255

# 数据增强
datagen = ImageDataGenerator(
    width_shift_range=0.1,
    height_shift_range=0.1,
    horizontal_flip=True,
    )

# 建立模型
model = se_resnet18(input_shape=(32,32,3), num_classes=10)

# 编译模型
model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

# 训练模型
history = model.fit_generator(datagen.flow(x_train, y_train, batch_size=128),
                              epochs=50,
                              validation_data=(x_test, y_test),
                              verbose=1)

# 打印最终结果
scores = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])

通过以上代码，我们可以得到模型的最终准确率。