LeNet

1. #coding=utf-8
2. from keras.models import Sequential  
3. from keras.layers import Dense,Flatten  
4. from keras.layers.convolutional import Conv2D,MaxPooling2D  
5. from keras.utils.np_utils import to_categorical  
6. import cPickle  
7. import gzip  
8. import numpy as np  
9. seed = 7
10. np.random.seed(seed)  
11. data = gzip.open(r'/media/wmy/document/BigData/kaggle/Digit Recognizer/mnist.pkl.gz')  
12. train_set,valid_set,test_set = cPickle.load(data)  
13. #train_x is [0,1]
14. train_x = train_set[0].reshape((-1,28,28,1))  
15. train_y = to_categorical(train_set[1])  
16. valid_x = valid_set[0].reshape((-1,28,28,1))  
17. valid_y = to_categorical(valid_set[1
    ])  
18. test_x = test_set[0].reshape((-1,28,28,1))  
19. test_y = to_categorical(test_set[1])  
20. model = Sequential()  
21. model.add(Conv2D(32,(5,5),strides=(1,1),input_shape=(28,28,1),padding='valid',activation='relu',kernel_initializer='uniform'))  
22. model.add(MaxPooling2D(pool_size=(2,2)))  
23. model.add(Conv2D(64
    ,(5,5),strides=(1,1),padding='valid',activation='relu',kernel_initializer='uniform'))  
24. model.add(MaxPooling2D(pool_size=(2,2)))  
25. model.add(Flatten())  
26. model.add(Dense(100,activation='relu'))  
27. model.add(Dense(10,activation='softmax'))  
28. model.compile(optimizer='sgd',loss='categorical_crossentropy',metrics=['accuracy'])  
29. model.summary()  
30. model.fit(train_x,train_y,validation_data=(valid_x,valid_y),batch_size=20,epochs=20,verbose=2)  
31. #[0.031825309940411217, 0.98979999780654904]
32. print model.evaluate(test_x,test_y,batch_size=20,verbose=2)  

AlexNet

1. #coding=utf-8
2. from keras.models import Sequential  
3. from keras.layers import Dense,Flatten,Dropout  
4. from keras.layers.convolutional import Conv2D,MaxPooling2D  
5. from keras.utils.np_utils import to_categorical  
6. import numpy as np  
7. seed = 7
8. np.random.seed(seed)  
9. model = Sequential()  
10. model.add(Conv2D(96,(11,11),strides=(4,4),input_shape=(227,227,3),padding='valid',activation='relu',kernel_initializer='uniform'))  
11. model.add(MaxPooling2D(pool_size=(3,3),strides=(2,2)))  
12. model.add(Conv2D(256,(5,5),strides=(1,1),padding='same',activation='relu',kernel_initializer='uniform'))  
13. model.add(MaxPooling2D(pool_size=(3,3),strides=(2,2)))  
14. model.add(Conv2D(384,(3,3),strides=(1,1),padding='same',activation='relu',kernel_initializer='uniform'))  
15. model.add(Conv2D(384,(3,3),strides=(1,1),padding='same',activation='relu',kernel_initializer='uniform'))  
16. model.add(Conv2D(256,(3,3),strides=(1,1),padding='same',activation='relu',kernel_initializer='uniform'))  
17. model.add(MaxPooling2D(pool_size=(3,3),strides=(2,2)))  
18. model.add(Flatten())  
19. model.add(Dense(4096,activation='relu'))  
20. model.add(Dropout(0.5))  
21. model.add(Dense(4096,activation='relu'))  
22. model.add(Dropout(0.5))  
23. model.add(Dense(1000,activation='softmax'))  
24. model.compile(loss='categorical_crossentropy',optimizer='sgd',metrics=['accuracy'])  
25. model.summary()  

ZFNet

1. #coding=utf-8
2. from keras.models import Sequential  
3. from keras.layers import Dense,Flatten,Dropout  
4. from keras.layers.convolutional import Conv2D,MaxPooling2D  
5. from keras.utils.np_utils import to_categorical  
6. import numpy as np  
7. seed = 7
8. np.random.seed(seed)  
9. model = Sequential()  
10. model.add(Conv2D(96,(7,7),strides=(2,2),input_shape=(224,224,3),padding='valid',activation='relu',kernel_initializer='uniform'))  
11. model.add(MaxPooling2D(pool_size=(3,3),strides=(2,2)))  
12. model.add(Conv2D(256,(5,5),strides=(2,2),padding='same',activation='relu',kernel_initializer='uniform'))  
13. model.add(MaxPooling2D(pool_size=(3,3),str