梧州市建设局官方网站,小工程承包,国内视频网站域名,付费资料网站开发项目要点
模型创建: model Sequential()添加卷积层: model.add(Dense(32, activationrelu, input_dim100)) # 第一层需要 input_dim添加dropout: model.add(Dropout(0.2))添加第二次网络: model.add(Dense(512, activationrelu)) # 除了first, 其他层不要输入shape添加输出…项目要点
模型创建: model Sequential()添加卷积层: model.add(Dense(32, activationrelu, input_dim100)) # 第一层需要 input_dim添加dropout: model.add(Dropout(0.2))添加第二次网络: model.add(Dense(512, activationrelu)) # 除了first, 其他层不要输入shape添加输出层: model.add(Dense(num_classes, activationsoftmax)) # last 通常使用softmaxTensorFlow 中使用 model.compile 方法来选择优化器和损失函数: optimizer: 优化器: 主要有: tf.train.AdamOptimizer , tf.train.RMSPropOptimizer , or tf.train.GradientDescentOptimizer . loss: 损失函数: 主要有:mean square error (mse, 回归), categorical_crossentropy (多分类) , and binary_crossentropy (二分类). metrics: 算法的评估标准, 一般分类用accuracy. model.fit(x_train, y_train, batch_size 64, epochs 20, validation_data (x_test, y_test)) # 模型训练score model.evaluate(x_test, y_test, verbose0) 两个返回值: [ 损失率 , 准确率 ]1 实例演示Keras的使用 (手写数字识别)
1.1 导包
import keras
from keras.datasets import mnist
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.optimizers import rmsprop_v2
1.2 导入数据
# 导入手写数字数据集
(x_train, y_train), (x_test, y_test) mnist.load_data()
print(x_train.shape, y_train.shape, x_test.shape, y_test.shape)
(60000, 28, 28) (60000,) (10000, 28, 28) (10000,)
import matplotlib.pyplot as plt
plt.imshow(x_train[0], cmap gray) 1.3 数据初步处理
# 对数据进行初步处理
x_train x_train.reshape(60000, 784)
x_test x_test.reshape(10000, 784)
x_train x_train.astype(float32)
x_test x_test.astype(float32)
x_train / 255
x_test / 255
print(x_train.shape, train samples) # (60000, 784) train samples
print(x_test.shape, test samples) # (10000, 784) test samples
1.4 数据初步处理
独热编码
import tensorflow
# 将标记结果转化为独热编码
num_classes 10
y_train tensorflow.keras.utils.to_categorical(y_train, num_classes)
y_test tensorflow.keras.utils.to_categorical(y_test, num_classes)
y_train 1.5 创建模型
# 创建顺序模型
model Sequential()
# 添加第一层网络, 512个神经元, 激活函数为relu
model.add(Dense(512, activationrelu, input_shape(784,)))
# 添加Dropout
model.add(Dropout(0.2))
# 第二层网络
model.add(Dense(512, activationrelu))
model.add(Dropout(0.2))
# 输出层
model.add(Dense(num_classes, activationsoftmax))
# 打印神经网络参数情况
model.summary() 1.6 模型训练
# 编译
model.compile(losscategorical_crossentropy,optimizerrmsprop,metrics[accuracy])batch_size 128
epochs 20
# 训练并打印中间过程
history model.fit(x_train, y_train,batch_sizebatch_size,epochsepochs,verbose1,validation_data(x_test, y_test))
# 计算预测数据的准确率
score model.evaluate(x_test, y_test, verbose0)
print(Test loss:, score[0]) # Test loss: 0.14742641150951385
print(Test accuracy:, score[1]) # Test accuracy: 0.9815000295639038
