目錄
1. TFlearn
1.1 TFlearn 安裝
pip3 安裝 TFlearn
pip3 install tflearn --user
Installing collected packages: tflearn
Successfully installed tflearn-0.3.2
1.2 官方例子
在這個例子中我們將對泰坦尼克號上的乘客進行存活可能性預測。
1.2.1 數據集加載
數據集中，每一個乘客的相關信息如下：
VARIABLE DESCRIPTIONS:
survived Survived
(0 = No; 1 = Yes)
pclass Passenger Class
(1 = st; 2 = nd; 3 = rd)
name Name
sex Sex
age Age
sibsp Number of Siblings/Spouses Aboard
parch Number of Parents/Children Aboard
ticket Ticket Number
fare Passenger Fare
其中總共有9項，我們將其分為標簽（label）和輸入（data），令標簽為是否存活，存活為1，那么輸入包含8項，其中我們認為姓名以及船票的號碼（可以由票價直接體現）對於我們預測乘客的存活幾率是沒有什么用的，所以在預處理中，我們將其拋棄。
數據集被存儲為 csv 文件格式。csv ，全稱為 Comma-Separated Values ，即逗號分隔值，其文本以純文本形式存儲表格數據，我們可以使用文本編輯器或 excel 直接打開。先加載數據到內存中
使用 load_csv() 函數從csv文件中讀取數據，並轉為 python List 。其中 target_column 參數用於表示我們的標簽列 id ，該函數將返回一個元組：（data,labels）。然后按照我們前面說的，拋棄輸入中的姓名以及船票號碼字段，並將性別字段轉為數值，0 表示男性，1 表示女性。
1.2.2 構建神經網絡
TFLearn中采用Tensor進行運算，因此這里的net都是Tensor，與TensorFlow中一樣，我們也可以將其中的某一個部分用TensorFlow中的函數自己寫，從而實現一些TFLearn庫中沒有的功能。其中全連接層的W(weights_init)和b(bias_init)可以指定，不過默認為W：'truncated_normal'，b：'zeros'，此外，其中的 activation 參數默認為'linear'。
1.2.3 訓練
其中 tflearn.DNN 是TFLearn中提供的一個模型 wrapper，相當於我們將很多功能包裝起來，我們給它一個 net 結構，生成一個 model 對象，然后調用model對象的訓練、預測、存儲等功能，DNN類有三個屬性（成員變量）：trainer，predictor，session。在fit()函數中n_epoch=10表示整個訓練數據集將會用10遍，batch_size=16表示一次用16個數據計算參數的更新。
最后利用訓練得到的模型進行預測：
import numpy as np
import tflearn
# Download the Titanic dataset
from tflearn.datasets import titanic
titanic.download_dataset('titanic_dataset.csv')
# Load CSV file, indicate that the first column represents labels
from tflearn.data_utils import load_csv
data, labels = load_csv('titanic_dataset.csv', target_column=0,
categorical_labels=True, n_classes=2)
# Preprocessing function
def preprocess(data, columns_to_ignore):
# Sort by descending id and delete columns
for id in sorted(columns_to_ignore, reverse=True):
[r.pop(id) for r in data]
for i in range(len(data)):
# Converting 'sex' field to float (id is 1 after removing labels column)
data[i][1] = 1. if data[i][1] == 'female' else 0.
return np.array(data, dtype=np.float32)
# Ignore 'name' and 'ticket' columns (id 1 & 6 of data array)
to_ignore=[1, 6]
# Preprocess data
data = preprocess(data, to_ignore)
# Build neural network
net = tflearn.input_data(shape=[None, 6])
net = tflearn.fully_connected(net, 32)
net = tflearn.fully_connected(net, 32)
net = tflearn.fully_connected(net, 2, activation='softmax')
net = tflearn.regression(net)
# Define model
model = tflearn.DNN(net)
# Start training (apply gradient descent algorithm)
model.fit(data, labels, n_epoch=10, batch_size=16, show_metric=True)
# Let's create some data for DiCaprio and Winslet
dicaprio = [3, 'Jack Dawson', 'male', 19, 0, 0, 'N/A', 5.0000]
winslet = [1, 'Rose DeWitt Bukater', 'female', 17, 1, 2, 'N/A', 100.0000]
# Preprocess data
dicaprio, winslet = preprocess([dicaprio, winslet], to_ignore)
# Predict surviving chances (class 1 results)
pred = model.predict([dicaprio, winslet])
print("DiCaprio Surviving Rate:", pred[0][1])
print("Winslet Surviving Rate:", pred[1][1])
1.2.4 測試結果
Training samples: 1309
Validation samples: 0
--
successfully opened CUDA library libcublas.so.10.0 locally
Training Step: 82 | total loss: 0.65318 | time: 3.584s
| Adam | epoch: 001 | loss: 0.65318 - acc: 0.6781 -- iter: 1309/1309
--
Training Step: 164 | total loss: 0.63713 | time: 1.298s
| Adam | epoch: 002 | loss: 0.63713 - acc: 0.6687 -- iter: 1309/1309
--
Training Step: 246 | total loss: 0.55357 | time: 1.354s
| Adam | epoch: 003 | loss: 0.55357 - acc: 0.7219 -- iter: 1309/1309
--
Training Step: 328 | total loss: 0.56566 | time: 1.312s
| Adam | epoch: 004 | loss: 0.56566 - acc: 0.7091 -- iter: 1309/1309
--
Training Step: 410 | total loss: 0.48417 | time: 1.311s
| Adam | epoch: 005 | loss: 0.48417 - acc: 0.7854 -- iter: 1309/1309
--
Training Step: 492 | total loss: 0.56114 | time: 1.300s
| Adam | epoch: 006 | loss: 0.56114 - acc: 0.7463 -- iter: 1309/1309
--
Training Step: 574 | total loss: 0.51057 | time: 1.289s
| Adam | epoch: 007 | loss: 0.51057 - acc: 0.7988 -- iter: 1309/1309
--
Training Step: 656 | total loss: 0.56562 | time: 1.312s
| Adam | epoch: 008 | loss: 0.56562 - acc: 0.7551 -- iter: 1309/1309
--
Training Step: 738 | total loss: 0.52883 | time: 1.324s
| Adam | epoch: 009 | loss: 0.52883 - acc: 0.7654 -- iter: 1309/1309
--
Training Step: 820 | total loss: 0.50510 | time: 1.340s
| Adam | epoch: 010 | loss: 0.50510 - acc: 0.7687 -- iter: 1309/1309
--
DiCaprio Surviving Rate: 0.17452878
Winslet Surviving Rate: 0.938663
我們的模型完成訓練，總體准確率在 76.87％，這意味着它可以預測76％總乘客的正確結果（幸存與否）。
其中 Dicaprio 是男主角，Winslet 為女主角，可以看出預測還是比較准的。
2. Keras
- keras官方文檔
- keras官方中文文檔
- kerea github : Deep Learning for humans
- git ee 國內鏡像
- 一個面向初學者的，友好的Keras入門教程
- CIFAR 10 官方數據
掌握 keras 可以大幅提升對開發效率和網絡結構的理解。優點：
- 模塊化
- 極簡主義
- 易擴展性
2.1 安裝Keras
pip3 install keras --user
Successfully installed keras-2.2.4
安裝完成后，進入python3，檢查一下安裝成果，import keras時，下方提示using TensorFlow backend,就證明Keras安裝成功並使用TensorFlow作為backend。
import keras
Using TensorFlow backend.
ModuleNotFoundError: No module named 'numpy.core._multiarray_umath'
ImportError: numpy.core.multiarray failed to import
這里有一個小問題，需要升級numpy包
pip3 install -i https://pypi.tuna.tsinghua.edu.cn/simple --upgrade numpy --user
Successfully installed numpy-1.16.3
然后keras成功安裝
import keras
Using TensorFlow backend.
2.2 keras 的模型
keras 的核心數據是模型。模型是用來組織網絡層的方式。模型有兩種，一種叫 Sequential 模型，另一種叫 Model 模型 。 Sequential 模型是一系列網絡層按順序構成的棧，是單輸入單輸出的，層與層之間只有相鄰關系，是最簡單的一種模型。
Keras 是一個用 Python 編寫的高級神經網絡 API，它能夠以 TensorFlow, CNTK, 或者 Theano 作為后端運行。Keras 的開發重點是支持快速的實驗。能夠以最小的時延把你的想法轉換為實驗結果，是做好研究的關鍵。
如果你在以下情況下需要深度學習庫，請使用 Keras：
- 允許簡單而快速的原型設計（由於用戶友好，高度模塊化，可擴展性）。
- 同時支持卷積神經網絡和循環神經網絡，以及兩者的組合。
- 在 CPU 和 GPU 上無縫運行。
- 查看文檔，請訪問 Keras.io。
Keras 兼容的 Python 版本: Python 2.7-3.6。
指導原則
-
用戶友好。 Keras 是為人類而不是為機器設計的 API。它把用戶體驗放在首要和中心位置。Keras 遵循減少認知困難的最佳實踐：它提供一致且簡單的 API，將常見用例所需的用戶操作數量降至最低，並且在用戶錯誤時提供清晰和可操作的反饋。
-
模塊化。 模型被理解為由獨立的、完全可配置的模塊構成的序列或圖。這些模塊可以以盡可能少的限制組裝在一起。特別是神經網絡層、損失函數、優化器、初始化方法、激活函數、正則化方法，它們都是可以結合起來構建新模型的模塊。
-
易擴展性。 新的模塊是很容易添加的（作為新的類和函數），現有的模塊已經提供了充足的示例。由於能夠輕松地創建可以提高表現力的新模塊，Keras 更加適合高級研究。
-
基於 Python 實現。 Keras 沒有特定格式的單獨配置文件。模型定義在 Python 代碼中，這些代碼緊湊，易於調試，並且易於擴展。
2.3 官方例子 ： 使用 ResNet 模型對 CIFAIR 10 數據集分類
在 examples 目錄 中，你可以找到真實數據集的示例模型：
- CIFAR10 小圖片分類：具有實時數據增強的卷積神經網絡 (CNN)
- IMDB 電影評論情感分類：基於詞序列的 LSTM
- Reuters 新聞主題分類：多層感知器 (MLP)
- MNIST 手寫數字分類：MLP & CNN
- 基於 LSTM 的字符級文本生成
2.3.1 Keras數據集和預訓練模型目錄
Keras下載的數據集在以下目錄中：
- win10 ：
C:\Users\user_name\.keras\datasets
其中一般化user_name是Administrator - Linux中，對一般用戶，用戶主目錄是：
/home/user_name，對於root用戶，主目錄是：/root
Keras下載的預訓練模型在一下目錄中：
- root用戶在
/root/.keras/models - 一般用戶在
/home/user_name/.keras/models
2.3.2 例程源碼
# https://github.com/keras-team/keras/tree/master/examples/cifar10_cnn.py
"""
#Trains a ResNet on the CIFAR10 dataset.
ResNet v1:
[Deep Residual Learning for Image Recognition
](https://arxiv.org/pdf/1512.03385.pdf)
ResNet v2:
[Identity Mappings in Deep Residual Networks
](https://arxiv.org/pdf/1603.05027.pdf)
Model|n|200-epoch accuracy|Original paper accuracy |sec/epoch GTX1080Ti
:------------|--:|-------:|-----------------------:|---:
ResNet20 v1| 3| 92.16 %| 91.25 %|35
ResNet32 v1| 5| 92.46 %| 92.49 %|50
ResNet44 v1| 7| 92.50 %| 92.83 %|70
ResNet56 v1| 9| 92.71 %| 93.03 %|90
ResNet110 v1| 18| 92.65 %| 93.39+-.16 %|165
ResNet164 v1| 27| - %| 94.07 %| -
ResNet1001 v1|N/A| - %| 92.39 %| -
Model|n|200-epoch accuracy|Original paper accuracy |sec/epoch GTX1080Ti
:------------|--:|-------:|-----------------------:|---:
ResNet20 v2| 2| - %| - %|---
ResNet32 v2|N/A| NA %| NA %| NA
ResNet44 v2|N/A| NA %| NA %| NA
ResNet56 v2| 6| 93.01 %| NA %|100
ResNet110 v2| 12| 93.15 %| 93.63 %|180
ResNet164 v2| 18| - %| 94.54 %| -
ResNet1001 v2|111| - %| 95.08+-.14 %| -
"""
# %matplotlib inline
# %config InlineBackend.figure_format = 'svg'
# calculate time using
import timeit
start = timeit.default_timer()
import keras
from keras.layers import Dense, Conv2D, BatchNormalization, Activation
from keras.layers import AveragePooling2D, Input, Flatten
from keras.optimizers import Adam
from keras.callbacks import ModelCheckpoint, LearningRateScheduler
from keras.callbacks import ReduceLROnPlateau
from keras.preprocessing.image import ImageDataGenerator
from keras.regularizers import l2
from keras import backend as K
from keras.models import Model
from keras.datasets import cifar10
import numpy as np
import os
# calculate time using
import timeit
start = timeit.default_timer()
# Training parameters
batch_size = 32 # orig paper trained all networks with batch_size=128
epochs = 10
data_augmentation = True
num_classes = 10
# Subtracting pixel mean improves accuracy
subtract_pixel_mean = True
# Model parameter
# ----------------------------------------------------------------------------
# | | 200-epoch | Orig Paper| 200-epoch | Orig Paper| sec/epoch
# Model | n | ResNet v1 | ResNet v1 | ResNet v2 | ResNet v2 | GTX1080Ti
# |v1(v2)| %Accuracy | %Accuracy | %Accuracy | %Accuracy | v1 (v2)
# ----------------------------------------------------------------------------
# ResNet20 | 3 (2)| 92.16 | 91.25 | ----- | ----- | 35 (---)
# ResNet32 | 5(NA)| 92.46 | 92.49 | NA | NA | 50 ( NA)
# ResNet44 | 7(NA)| 92.50 | 92.83 | NA | NA | 70 ( NA)
# ResNet56 | 9 (6)| 92.71 | 93.03 | 93.01 | NA | 90 (100)
# ResNet110 |18(12)| 92.65 | 93.39+-.16| 93.15 | 93.63 | 165(180)
# ResNet164 |27(18)| ----- | 94.07 | ----- | 94.54 | ---(---)
# ResNet1001| (111)| ----- | 92.39 | ----- | 95.08+-.14| ---(---)
# ---------------------------------------------------------------------------
n = 3
# Model version
# Orig paper: version = 1 (ResNet v1), Improved ResNet: version = 2 (ResNet v2)
version = 1
# Computed depth from supplied model parameter n
if version == 1:
depth = n * 6 + 2
elif version == 2:
depth = n * 9 + 2
# Model name, depth and version
model_type = 'ResNet%dv%d' % (depth, version)
# Load the CIFAR10 data.
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
# Input image dimensions.
input_shape = x_train.shape[1:]
# Normalize data.
x_train = x_train.astype('float32') / 255
x_test = x_test.astype('float32') / 255
# If subtract pixel mean is enabled
if subtract_pixel_mean:
x_train_mean = np.mean(x_train, axis=0)
x_train -= x_train_mean
x_test -= x_train_mean
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
print('y_train shape:', y_train.shape)
# 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)
def lr_schedule(epoch):
"""Learning Rate Schedule
Learning rate is scheduled to be reduced after 80, 120, 160, 180 epochs.
Called automatically every epoch as part of callbacks during training.
# Arguments
epoch (int): The number of epochs
# Returns
lr (float32): learning rate
"""
lr = 1e-3
if epoch > 180:
lr *= 0.5e-3
elif epoch > 160:
lr *= 1e-3
elif epoch > 120:
lr *= 1e-2
elif epoch > 80:
lr *= 1e-1
print('Learning rate: ', lr)
return lr
def resnet_layer(inputs,
num_filters=16,
kernel_size=3,
strides=1,
activation='relu',
batch_normalization=True,
conv_first=True):
"""2D Convolution-Batch Normalization-Activation stack builder
# Arguments
inputs (tensor): input tensor from input image or previous layer
num_filters (int): Conv2D number of filters
kernel_size (int): Conv2D square kernel dimensions
strides (int): Conv2D square stride dimensions
activation (string): activation name
batch_normalization (bool): whether to include batch normalization
conv_first (bool): conv-bn-activation (True) or
bn-activation-conv (False)
# Returns
x (tensor): tensor as input to the next layer
"""
conv = Conv2D(num_filters,
kernel_size=kernel_size,
strides=strides,
padding='same',
kernel_initializer='he_normal',
kernel_regularizer=l2(1e-4))
x = inputs
if conv_first:
x = conv(x)
if batch_normalization:
x = BatchNormalization()(x)
if activation is not None:
x = Activation(activation)(x)
else:
if batch_normalization:
x = BatchNormalization()(x)
if activation is not None:
x = Activation(activation)(x)
x = conv(x)
return x
def resnet_v1(input_shape, depth, num_classes=10):
"""ResNet Version 1 Model builder [a]
Stacks of 2 x (3 x 3) Conv2D-BN-ReLU
Last ReLU is after the shortcut connection.
At the beginning of each stage, the feature map size is halved (downsampled)
by a convolutional layer with strides=2, while the number of filters is
doubled. Within each stage, the layers have the same number filters and the
same number of filters.
Features maps sizes:
stage 0: 32x32, 16
stage 1: 16x16, 32
stage 2: 8x8, 64
The Number of parameters is approx the same as Table 6 of [a]:
ResNet20 0.27M
ResNet32 0.46M
ResNet44 0.66M
ResNet56 0.85M
ResNet110 1.7M
# Arguments
input_shape (tensor): shape of input image tensor
depth (int): number of core convolutional layers
num_classes (int): number of classes (CIFAR10 has 10)
# Returns
model (Model): Keras model instance
"""
if (depth - 2) % 6 != 0:
raise ValueError('depth should be 6n+2 (eg 20, 32, 44 in [a])')
# Start model definition.
num_filters = 16
num_res_blocks = int((depth - 2) / 6)
inputs = Input(shape=input_shape)
x = resnet_layer(inputs=inputs)
# Instantiate the stack of residual units
for stack in range(3):
for res_block in range(num_res_blocks):
strides = 1
if stack > 0 and res_block == 0: # first layer but not first stack
strides = 2 # downsample
y = resnet_layer(inputs=x,
num_filters=num_filters,
strides=strides)
y = resnet_layer(inputs=y,
num_filters=num_filters,
activation=None)
if stack > 0 and res_block == 0: # first layer but not first stack
# linear projection residual shortcut connection to match
# changed dims
x = resnet_layer(inputs=x,
num_filters=num_filters,
kernel_size=1,
strides=strides,
activation=None,
batch_normalization=False)
x = keras.layers.add([x, y])
x = Activation('relu')(x)
num_filters *= 2
# Add classifier on top.
# v1 does not use BN after last shortcut connection-ReLU
x = AveragePooling2D(pool_size=8)(x)
y = Flatten()(x)
outputs = Dense(num_classes,
activation='softmax',
kernel_initializer='he_normal')(y)
# Instantiate model.
model = Model(inputs=inputs, outputs=outputs)
return model
def resnet_v2(input_shape, depth, num_classes=10):
"""ResNet Version 2 Model builder [b]
Stacks of (1 x 1)-(3 x 3)-(1 x 1) BN-ReLU-Conv2D or also known as
bottleneck layer
First shortcut connection per layer is 1 x 1 Conv2D.
Second and onwards shortcut connection is identity.
At the beginning of each stage, the feature map size is halved (downsampled)
by a convolutional layer with strides=2, while the number of filter maps is
doubled. Within each stage, the layers have the same number filters and the
same filter map sizes.
Features maps sizes:
conv1 : 32x32, 16
stage 0: 32x32, 64
stage 1: 16x16, 128
stage 2: 8x8, 256
# Arguments
input_shape (tensor): shape of input image tensor
depth (int): number of core convolutional layers
num_classes (int): number of classes (CIFAR10 has 10)
# Returns
model (Model): Keras model instance
"""
if (depth - 2) % 9 != 0:
raise ValueError('depth should be 9n+2 (eg 56 or 110 in [b])')
# Start model definition.
num_filters_in = 16
num_res_blocks = int((depth - 2) / 9)
inputs = Input(shape=input_shape)
# v2 performs Conv2D with BN-ReLU on input before splitting into 2 paths
x = resnet_layer(inputs=inputs,
num_filters=num_filters_in,
conv_first=True)
# Instantiate the stack of residual units
for stage in range(3):
for res_block in range(num_res_blocks):
activation = 'relu'
batch_normalization = True
strides = 1
if stage == 0:
num_filters_out = num_filters_in * 4
if res_block == 0: # first layer and first stage
activation = None
batch_normalization = False
else:
num_filters_out = num_filters_in * 2
if res_block == 0: # first layer but not first stage
strides = 2 # downsample
# bottleneck residual unit
y = resnet_layer(inputs=x,
num_filters=num_filters_in,
kernel_size=1,
strides=strides,
activation=activation,
batch_normalization=batch_normalization,
conv_first=False)
y = resnet_layer(inputs=y,
num_filters=num_filters_in,
conv_first=False)
y = resnet_layer(inputs=y,
num_filters=num_filters_out,
kernel_size=1,
conv_first=False)
if res_block == 0:
# linear projection residual shortcut connection to match
# changed dims
x = resnet_layer(inputs=x,
num_filters=num_filters_out,
kernel_size=1,
strides=strides,
activation=None,
batch_normalization=False)
x = keras.layers.add([x, y])
num_filters_in = num_filters_out
# Add classifier on top.
# v2 has BN-ReLU before Pooling
x = BatchNormalization()(x)
x = Activation('relu')(x)
x = AveragePooling2D(pool_size=8)(x)
y = Flatten()(x)
outputs = Dense(num_classes,
activation='softmax',
kernel_initializer='he_normal')(y)
# Instantiate model.
model = Model(inputs=inputs, outputs=outputs)
return model
if version == 2:
model = resnet_v2(input_shape=input_shape, depth=depth)
else:
model = resnet_v1(input_shape=input_shape, depth=depth)
model.compile(loss='categorical_crossentropy',
optimizer=Adam(lr=lr_schedule(0)),
metrics=['accuracy'])
model.summary()
print(model_type)
# Prepare model model saving directory.
save_dir = os.path.join(os.getcwd(), 'saved_models')
model_name = 'cifar10_%s_model.{epoch:03d}.h5' % model_type
if not os.path.isdir(save_dir):
os.makedirs(save_dir)
filepath = os.path.join(save_dir, model_name)
# Prepare callbacks for model saving and for learning rate adjustment.
checkpoint = ModelCheckpoint(filepath=filepath,
monitor='val_acc',
verbose=1,
save_best_only=True)
lr_scheduler = LearningRateScheduler(lr_schedule)
lr_reducer = ReduceLROnPlateau(factor=np.sqrt(0.1),
cooldown=0,
patience=5,
min_lr=0.5e-6)
callbacks = [checkpoint, lr_reducer, lr_scheduler]
# Run training, with or without data augmentation.
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=callbacks)
else:
print('Using real-time data augmentation.')
# This will do preprocessing and realtime data augmentation:
datagen = ImageDataGenerator(
# set input mean to 0 over the dataset
featurewise_center=False,
# set each sample mean to 0
samplewise_center=False,
# divide inputs by std of dataset
featurewise_std_normalization=False,
# divide each input by its std
samplewise_std_normalization=False,
# apply ZCA whitening
zca_whitening=False,
# epsilon for ZCA whitening
zca_epsilon=1e-06,
# randomly rotate images in the range (deg 0 to 180)
rotation_range=0,
# randomly shift images horizontally
width_shift_range=0.1,
# randomly shift images vertically
height_shift_range=0.1,
# set range for random shear
shear_range=0.,
# set range for random zoom
zoom_range=0.,
# set range for random channel shifts
channel_shift_range=0.,
# set mode for filling points outside the input boundaries
fill_mode='nearest',
# value used for fill_mode = "constant"
cval=0.,
# randomly flip images
horizontal_flip=True,
# randomly flip images
vertical_flip=False,
# set rescaling factor (applied before any other transformation)
rescale=None,
# set function that will be applied on each input
preprocessing_function=None,
# image data format, either "channels_first" or "channels_last"
data_format=None,
# fraction of images reserved for validation (strictly between 0 and 1)
validation_split=0.0)
# Compute quantities required for featurewise normalization
# (std, mean, and principal components if ZCA whitening is applied).
datagen.fit(x_train)
# Fit the model on the batches generated by datagen.flow().
model.fit_generator(datagen.flow(x_train, y_train, batch_size=batch_size),
steps_per_epoch=x_train.shape[0],
validation_data=(x_test, y_test),
epochs=epochs, verbose=1, workers=4,
callbacks=callbacks)
# Score trained model.
scores = model.evaluate(x_test, y_test, verbose=1)
print('Test loss:', scores[0])
print('Test accuracy:', scores[1])
# output time using
end = timeit.default_timer()
tdf = end -start
timeh = tdf // 3600
timem = tdf // 60
times = tdf % 60
print("use time: " , int(timeh) , "h" , int(timem) , "m" ,times, "s")
# output time using
end = timeit.default_timer()
tdf = end -start
timeh = tdf // 3600
timem = tdf // 60
times = tdf % 60
print("use time: " , int(timeh) , "h" , int(timem) , "m" ,times, "s")
2.3.3 訓練
直接運行后會有錯誤
python3 cifar10_cnn.py
ValueError: steps_per_epoch=None is only valid for a generator based on the keras.utils.Sequence class. Please specify steps_per_epoch or use the keras.utils.Sequence class.
這個是由於版本更迭，有些函數的參數作了修改
只需要將 cifar10_resnet.py 文件中
model.fit_generator(datagen.flow(x_train, y_train, batch_size=batch_size),
validation_data=(x_test, y_test),
epochs=epochs, verbose=1, workers=4,
callbacks=callbacks)
修改為
model.fit_generator(datagen.flow(x_train, y_train, batch_size=batch_size),
steps_per_epoch=x_train.shape[0] // batch_size,
validation_data=(x_test, y_test),
epochs=epochs, verbose=1, workers=4,
callbacks=callbacks)
2.3.3 測試結果
Using TensorFlow backend.
x_train shape: (50000, 32, 32, 3)
50000 train samples
10000 test samples
y_train shape: (50000, 1)
Learning rate: 0.001
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) (None, 32, 32, 3) 0
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 32, 32, 16) 448 input_1[0][0]
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 32, 32, 16) 64 conv2d_1[0][0]
__________________________________________________________________________________________________
activation_1 (Activation) (None, 32, 32, 16) 0 batch_normalization_1[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 32, 32, 16) 2320 activation_1[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 32, 32, 16) 64 conv2d_2[0][0]
__________________________________________________________________________________________________
activation_2 (Activation) (None, 32, 32, 16) 0 batch_normalization_2[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 32, 32, 16) 2320 activation_2[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 32, 32, 16) 64 conv2d_3[0][0]
__________________________________________________________________________________________________
add_1 (Add) (None, 32, 32, 16) 0 activation_1[0][0]
batch_normalization_3[0][0]
__________________________________________________________________________________________________
activation_3 (Activation) (None, 32, 32, 16) 0 add_1[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 32, 32, 16) 2320 activation_3[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 32, 32, 16) 64 conv2d_4[0][0]
__________________________________________________________________________________________________
activation_4 (Activation) (None, 32, 32, 16) 0 batch_normalization_4[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 32, 32, 16) 2320 activation_4[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 32, 32, 16) 64 conv2d_5[0][0]
__________________________________________________________________________________________________
add_2 (Add) (None, 32, 32, 16) 0 activation_3[0][0]
batch_normalization_5[0][0]
__________________________________________________________________________________________________
activation_5 (Activation) (None, 32, 32, 16) 0 add_2[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 32, 32, 16) 2320 activation_5[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 32, 32, 16) 64 conv2d_6[0][0]
__________________________________________________________________________________________________
activation_6 (Activation) (None, 32, 32, 16) 0 batch_normalization_6[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 32, 32, 16) 2320 activation_6[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 32, 32, 16) 64 conv2d_7[0][0]
__________________________________________________________________________________________________
add_3 (Add) (None, 32, 32, 16) 0 activation_5[0][0]
batch_normalization_7[0][0]
__________________________________________________________________________________________________
activation_7 (Activation) (None, 32, 32, 16) 0 add_3[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 16, 16, 32) 4640 activation_7[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 16, 16, 32) 128 conv2d_8[0][0]
__________________________________________________________________________________________________
activation_8 (Activation) (None, 16, 16, 32) 0 batch_normalization_8[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 16, 16, 32) 9248 activation_8[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 16, 16, 32) 544 activation_7[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 16, 16, 32) 128 conv2d_9[0][0]
__________________________________________________________________________________________________
add_4 (Add) (None, 16, 16, 32) 0 conv2d_10[0][0]
batch_normalization_9[0][0]
__________________________________________________________________________________________________
activation_9 (Activation) (None, 16, 16, 32) 0 add_4[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 16, 16, 32) 9248 activation_9[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 16, 16, 32) 128 conv2d_11[0][0]
__________________________________________________________________________________________________
activation_10 (Activation) (None, 16, 16, 32) 0 batch_normalization_10[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 16, 16, 32) 9248 activation_10[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 16, 16, 32) 128 conv2d_12[0][0]
__________________________________________________________________________________________________
add_5 (Add) (None, 16, 16, 32) 0 activation_9[0][0]
batch_normalization_11[0][0]
__________________________________________________________________________________________________
activation_11 (Activation) (None, 16, 16, 32) 0 add_5[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 16, 16, 32) 9248 activation_11[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 16, 16, 32) 128 conv2d_13[0][0]
__________________________________________________________________________________________________
activation_12 (Activation) (None, 16, 16, 32) 0 batch_normalization_12[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 16, 16, 32) 9248 activation_12[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 16, 16, 32) 128 conv2d_14[0][0]
__________________________________________________________________________________________________
add_6 (Add) (None, 16, 16, 32) 0 activation_11[0][0]
batch_normalization_13[0][0]
__________________________________________________________________________________________________
activation_13 (Activation) (None, 16, 16, 32) 0 add_6[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 8, 8, 64) 18496 activation_13[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 8, 8, 64) 256 conv2d_15[0][0]
__________________________________________________________________________________________________
activation_14 (Activation) (None, 8, 8, 64) 0 batch_normalization_14[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 8, 8, 64) 36928 activation_14[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 8, 8, 64) 2112 activation_13[0][0]
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 8, 8, 64) 256 conv2d_16[0][0]
__________________________________________________________________________________________________
add_7 (Add) (None, 8, 8, 64) 0 conv2d_17[0][0]
batch_normalization_15[0][0]
__________________________________________________________________________________________________
activation_15 (Activation) (None, 8, 8, 64) 0 add_7[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 8, 8, 64) 36928 activation_15[0][0]
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 8, 8, 64) 256 conv2d_18[0][0]
__________________________________________________________________________________________________
activation_16 (Activation) (None, 8, 8, 64) 0 batch_normalization_16[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 8, 8, 64) 36928 activation_16[0][0]
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 8, 8, 64) 256 conv2d_19[0][0]
__________________________________________________________________________________________________
add_8 (Add) (None, 8, 8, 64) 0 activation_15[0][0]
batch_normalization_17[0][0]
__________________________________________________________________________________________________
activation_17 (Activation) (None, 8, 8, 64) 0 add_8[0][0]
__________________________________________________________________________________________________
conv2d_20 (Conv2D) (None, 8, 8, 64) 36928 activation_17[0][0]
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 8, 8, 64) 256 conv2d_20[0][0]
__________________________________________________________________________________________________
activation_18 (Activation) (None, 8, 8, 64) 0 batch_normalization_18[0][0]
__________________________________________________________________________________________________
conv2d_21 (Conv2D) (None, 8, 8, 64) 36928 activation_18[0][0]
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 8, 8, 64) 256 conv2d_21[0][0]
__________________________________________________________________________________________________
add_9 (Add) (None, 8, 8, 64) 0 activation_17[0][0]
batch_normalization_19[0][0]
__________________________________________________________________________________________________
activation_19 (Activation) (None, 8, 8, 64) 0 add_9[0][0]
__________________________________________________________________________________________________
average_pooling2d_1 (AveragePoo (None, 1, 1, 64) 0 activation_19[0][0]
__________________________________________________________________________________________________
flatten_1 (Flatten) (None, 64) 0 average_pooling2d_1[0][0]
__________________________________________________________________________________________________
dense_1 (Dense) (None, 10) 650 flatten_1[0][0]
==================================================================================================
Total params: 274,442
Trainable params: 273,066
Non-trainable params: 1,376
__________________________________________________________________________________________________
ResNet20v1
Using real-time data augmentation.
Epoch 1/10
Learning rate: 0.001
successfully opened CUDA library libcublas.so.10.0 locally
50000/50000 [==============================] - 11286s 226ms/step - loss: 0.7185 - acc: 0.8164 - val_loss: 0.7312 - val_acc: 0.8302
訓練一個 Epoch 需要3個小時左右，訓練后測試集精度為83.03%。例子需要訓練200個Epoch，Jentson Nano 的 0.5T 的算力太差，不適合訓練模型，計算量太大選擇放棄。