def init(self, Client_num, iid_data, epoch, batch_size):
self.client_num = Client_num
self.epochs = epoch
self.batch_size = batch_size
if (iid_data == 1):
self.images_train_split, self.labels_train_split = self.data.iid_shards(
Client_num)
elif (iid_data == 0):
self.images_train_split, self.labels_train_split = self.data.noniid(
Client_num, 2000)
# self.images_train_split=self.images_train_split.astype('float32')
# self.data.x_test=self.data.x_test.astype('float32')
self.data.y_test = to_categorical(self.data.y_test, self.num_classes)
self.model = LSUVinit(
self.model, self.images_train_split[0][:batch_size].reshape(
batch_size, 32, 32, 3))
def LSUV_init(self,
train_batch,
batch_size=None,
modelname='model_norm',
postfix='',
sub_modelname='main_seq'):
modelname = modelname + postfix
if not batch_size:
batch_size = self.BATCH_SIZE
LSUVinit(self.models[modelname].get_layer(sub_modelname),
train_batch,
batch_size=batch_size)
class FitNet4:
history = LossHistory()
batch_size = 0
num_classes = 10
epochs = 0
client_num = 0
Input_Shape = (32, 32, 3)
#data_augmentation = True
data_augmentation = False
# The data, shuffled and split between train and test sets:
#(x_train, y_train), (x_test, y_test) = cifar10.load_data()
data = Cifar10_shards()
# print('x_train shape:', x_train.shape)
# print(x_train.shape[0], 'train samples')
# print(x_test.shape[0], 'test samples')
# Convert class vectors to binary class matrices.
# y_train = keras.utils.to_categorical(y_train, num_classes)
# y_test = keras.utils.to_categorical(y_test, num_classes)
model = Sequential()
model.add(Conv2D(32, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(32, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(32, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(48, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(48, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(80, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(80, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(80, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(80, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(80, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Conv2D(128, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(128, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(128, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(128, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(Conv2D(128, (3, 3), padding='same', input_shape=Input_Shape))
model.add(Activation('relu'))
model.add(GlobalMaxPooling2D())
model.add(Dropout(0.25))
#model.add(ZeroPadding2D((1, 1)))
#model.add(MaxPooling2D(pool_size=(2, 2)))
#model.add(Flatten())
#model.add(Dropout(0.2))
'''
model.add(Conv2D(512, (3, 3), padding='same'))
model.add(Activation('relu'))
model.add(ZeroPadding2D((1, 1)))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
'''
model.add(Dense(500))
model.add(Activation('relu'))
model.add(Dropout(0.25))
model.add(Dense(num_classes))
model.add(Activation('softmax'))
# initiate RMSprop optimizer
opt = keras.optimizers.Adam(lr=0.0001)
# Let's train the model using RMSprop
model.compile(loss='categorical_crossentropy',
optimizer=opt,
metrics=['accuracy'])
def init(self, Client_num, iid_data, epoch, batch_size):
self.client_num = Client_num
self.epochs = epoch
self.batch_size = batch_size
if (iid_data == 1):
self.images_train_split, self.labels_train_split = self.data.iid_shards(
Client_num)
elif (iid_data == 0):
self.images_train_split, self.labels_train_split = self.data.noniid(
Client_num, 2000)
# self.images_train_split=self.images_train_split.astype('float32')
# self.data.x_test=self.data.x_test.astype('float32')
self.data.y_test = to_categorical(self.data.y_test, self.num_classes)
self.model = LSUVinit(
self.model, self.images_train_split[0][:batch_size].reshape(
batch_size, 32, 32, 3))
# self.model = LSUVinit(self.model,self.images_train_split[0][:batch_size,:,:,:])
#self.tbCallBack = keras.callbacks.TensorBoard(log_dir='./Graph2', histogram_freq=0, write_graph=True, write_images=True)
# x_train = x_train.astype('float32')
# x_test = x_test.astype('float32')
def run(self):
Weight, Biases = [], []
init_weight, init_biases = self.get_parameter()
for c in range(self.client_num):
self.model.fit(
self.images_train_split[c].reshape(50000 // self.client_num,
32, 32, 3),
self.labels_train_split[c].reshape(50000 // self.client_num,
10),
batch_size=self.batch_size,
epochs=self.epochs,
validation_data=(self.data.x_test.reshape(10000, 32, 32, 3),
self.data.y_test.reshape(10000, 10)),
shuffle=True,
callbacks=[])
temp_weight, temp_biases = self.get_parameter()
Weight.append(temp_weight)
Biases.append(temp_biases)
self.set_parameter(init_weight, init_biases)
#print(Weight)
return [Weight, Biases]
def next(self, weight_vector, biases_vector):
Weight, Biases = [], []
self.set_parameter_from_vector(weight_vector, biases_vector)
init_weight, init_biases = self.get_parameter()
#print("init_weight")
# print(init_weight)
for c in range(self.client_num):
self.model.fit(
self.images_train_split[c].reshape(50000 // self.client_num,
32, 32, 3),
self.labels_train_split[c].reshape(50000 // self.client_num,
10),
batch_size=self.batch_size,
epochs=self.epochs,
validation_data=(self.data.x_test.reshape(10000, 32, 32, 3),
self.data.y_test.reshape(10000, 10)),
shuffle=True,
callbacks=[])
temp_weight, temp_biases = self.get_parameter()
Weight.append(temp_weight)
Biases.append(temp_biases)
self.set_parameter(init_weight, init_biases)
return [Weight, Biases]
def benchmark(self):
self.model.fit(self.images_train_split[0].reshape(50000, 32, 32, 3),
self.labels_train_split[0].reshape(50000, 10),
batch_size=self.batch_size,
epochs=self.epochs,
validation_data=(self.data.x_test.reshape(
10000, 32, 32,
3), self.data.y_test.reshape(10000, 10)),
shuffle=True,
callbacks=[self.history])
self.history.acc_plot('epoch')
def test(self, weight_vector, biases_vector):
self.set_parameter_from_vector(weight_vector, biases_vector)
acc = self.model.evaluate(self.data.x_test.reshape(10000, 32, 32, 3),
self.data.y_test.reshape(10000, 10))
return acc[1]
def get_parameter(self):
temp_weight, temp_biases = [], []
for layer in self.model.layers:
if isinstance(layer, (Dense, Conv2D)):
parameter = layer.get_weights()
temp_weight.append(parameter[0])
temp_biases.append(parameter[1])
return temp_weight, temp_biases
def set_parameter(self, weight, biases):
i = 0
for layer in self.model.layers:
if isinstance(layer, (Dense, Conv2D)):
layer.set_weights([weight[i], biases[i]])
i += 1
def set_parameter_from_vector(self, weight_vector, biases_vector):
#0-2 weight: 3*3*32=288 biases:32
#3-4 weight: 3*3*48=432 biases:48
#5-9 weight: 3*3*80=720 biases:80
#10-14 weight: 3*3*128=1152 biases:128
#15 weight: 128*500=64000 biases:500
#16 weight: 500*10=5000 biases:10
# print(weight_vector[0:10])
weight_vector = np.array(weight_vector)
biases_vector = np.array(biases_vector)
weight_vector /= self.client_num
biases_vector /= self.client_num
weight_pointer = 0
biases_pointer = 0
i = 0
for layer in self.model.layers:
if isinstance(layer, (Dense, Conv2D)):
if (i == 0):
# print(weight_pointer,weight_pointer)
# print(weight_vector[weight_pointer:weight_pointer+864])
weight = weight_vector[weight_pointer:weight_pointer +
864].reshape(3, 3, 3, 32)
weight_pointer += 864
# print(weight)
biases = biases_vector[biases_pointer:biases_pointer + 32]
biases_pointer += 32
elif (i <= 2):
weight = weight_vector[weight_pointer:weight_pointer +
9216].reshape(3, 3, 32, 32)
weight_pointer += 9216
biases = biases_vector[biases_pointer:biases_pointer + 32]
biases_pointer += 32
elif (i == 3):
weight = weight_vector[weight_pointer:weight_pointer +
13824].reshape(3, 3, 32, 48)
weight_pointer += 13824
biases = biases_vector[biases_pointer:biases_pointer + 48]
biases_pointer += 48
elif (i == 4):
weight = weight_vector[weight_pointer:weight_pointer +
20736].reshape(3, 3, 48, 48)
weight_pointer += 20736
biases = biases_vector[biases_pointer:biases_pointer + 48]
biases_pointer += 48
elif (i == 5):
weight = weight_vector[weight_pointer:weight_pointer +
34560].reshape(3, 3, 48, 80)
weight_pointer += 34560
biases = biases_vector[biases_pointer:biases_pointer + 80]
biases_pointer += 80
elif (i <= 9):
weight = weight_vector[weight_pointer:weight_pointer +
57600].reshape(3, 3, 80, 80)
weight_pointer += 57600
biases = biases_vector[biases_pointer:biases_pointer + 80]
biases_pointer += 80
elif (i == 10):
weight = weight_vector[weight_pointer:weight_pointer +
92160].reshape(3, 3, 80, 128)
weight_pointer += 92160
biases = biases_vector[biases_pointer:biases_pointer + 128]
biases_pointer += 128
elif (i <= 14):
weight = weight_vector[weight_pointer:weight_pointer +
147456].reshape(3, 3, 128, 128)
weight_pointer += 147456
biases = biases_vector[biases_pointer:biases_pointer + 128]
biases_pointer += 128
elif (i == 15):
weight = weight_vector[weight_pointer:weight_pointer +
64000].reshape(128, 500)
weight_pointer += 64000
biases = biases_vector[biases_pointer:biases_pointer + 500]
biases_pointer += 500
elif (i == 16):
weight = weight_vector[weight_pointer:weight_pointer +
5000].reshape(500, 10)
weight_pointer += 5000
biases = biases_vector[biases_pointer:biases_pointer + 10]
biases_pointer += 10
layer.set_weights([weight, biases])
i += 1
model.add(Dense(500))
model.add(Activation('relu'))
model.add(Dropout(0.25))
model.add(Dense(num_classes))
model.add(Activation('softmax'))
# initiate RMSprop optimizer
opt = keras.optimizers.Adam(lr=0.0001)
# Let's train the model using RMSprop
model.compile(loss='categorical_crossentropy',
optimizer=opt,
metrics=['accuracy'])
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
x_train /= 255
x_test /= 255
model = LSUVinit(model, x_train[:batch_size, :, :, :])
tbCallBack = keras.callbacks.TensorBoard(log_dir='./Graph2',
histogram_freq=0,
write_graph=True,
write_images=True)
if not data_augmentation:
print('Not using data augmentation.')
model.fit(x_train,
y_train,
batch_size=batch_size,
epochs=epochs,
validation_data=(x_test, y_test),
shuffle=True,
callbacks=[tbCallBack])
else:
print('Using real-time data augmentation.')