keras模拟alexnet， mnist test（第三次）（改变dropout的权值)（最高精度99.51）

本文链接：https://blog.youkuaiyun.com/weixin_42160791/article/details/88802292

本文介绍了一种使用Keras构建的深度学习模型，用于解决MNIST手写数字识别任务。模型采用卷积神经网络（CNN），包含多个卷积层、池化层、全连接层及Dropout层来提高泛化能力。通过对MNIST数据集进行预处理和训练，该模型能有效识别手写数字。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

from keras.models import Sequential  
from keras.layers import Input, Dense, Dropout, Activation, BatchNormalization
from keras.models import Model
from keras.optimizers import SGD
from keras.datasets import mnist
from keras.layers import Conv2D, MaxPooling2D, Flatten
import numpy as np


tBatchSize = 100
'''第一步：选择模型'''

nn = Sequential() # 采用贯序模型
'''第二步：构建网络层'''
nn.add(Conv2D(32,(3,3), input_shape=(28,28,1)))
nn.add(BatchNormalization())
nn.add(Activation('relu'))
nn.add(MaxPooling2D(pool_size=(2, 2),strides=1))

nn.add(Conv2D(64,(3,3)))
nn.add(BatchNormalization())
nn.add(Activation('relu'))
nn.add(MaxPooling2D(pool_size=(2, 2)))

nn.add(Conv2D(128,(3,3)))
nn.add(BatchNormalization())
nn.add(Activation('relu'))
nn.add(MaxPooling2D(pool_size=(2, 2)))

nn.add(Conv2D(256,(3,3)))
nn.add(BatchNormalization())
nn.add(Activation('relu'))


nn.add(MaxPooling2D(pool_size=(2, 2)))

nn.add(Flatten())

nn.add(Dense(500)) 
nn.add(BatchNormalization())
nn.add(Activation('relu')) 
nn.add(Dropout(0.6)) 
 

nn.add(Dense(500))
nn.add(BatchNormalization()) 
nn.add(Activation('relu'))
nn.add(Dropout(0.6))
 
nn.add(Dense(500)) 
nn.add(Activation('relu'))

nn.add(Dense(10))
nn.add(BatchNormalization())
nn.add(Activation('softmax'))


nn.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])     

'''第四步：训练'''
(X_train, y_train), (X_test, y_test) = mnist.load_data() 
 

X_train = X_train.reshape(X_train.shape[0], X_train.shape[1] , X_train.shape[2],1)
X_test = X_test.reshape(X_test.shape[0], X_test.shape[1] , X_test.shape[2],1)


Y_train = (np.arange(10) == y_train[:, None]).astype(int)  
Y_test = (np.arange(10) == y_test[:, None]).astype(int)    

X_train = X_train/256.0
X_test = X_test / 256.0

nn.fit(X_train, Y_train, batch_size=tBatchSize, epochs=1000, 
              shuffle=True, verbose=2, validation_split=0.3)