tensorflow2.0教程- Keras 快速入门

最新推荐文章于 2020-07-03 22:20:22 发布

qq_41483834

最新推荐文章于 2020-07-03 22:20:22 发布

阅读量194

点赞数

分类专栏： ocr

原文链接：https://blog.csdn.net/qq_31456593/article/details/88377117

版权

ocr 专栏收录该内容

5 篇文章 0 订阅

订阅专栏

本文链接： https://blog.csdn.net/qq_31456593/article/details/88377117

tensorflow2.0教程-tensorflow.keras 快速入门

Tensorflow 2.0 教程持续更新： https://blog.csdn.net/qq_31456593/article/details/88606284

TensorFlow 2.0 教程- Keras 快速入门
 TensorFlow 2.0 教程-keras 函数api
TensorFlow 2.0 教程-使用keras训练模型
 TensorFlow 2.0 教程-用keras构建自己的网络层
 TensorFlow 2.0 教程-keras模型保存和序列化
 TensorFlow 2.0 教程-eager模式
 TensorFlow 2.0 教程-Variables
TensorFlow 2.0 教程–AutoGraph

TensorFlow 2.0 深度学习实践

TensorFlow2.0 教程-图像分类
 TensorFlow2.0 教程-文本分类
 TensorFlow2.0 教程-过拟合和欠拟合

完整tensorflow2.0教程代码请看tensorflow2.0：中文教程tensorflow2_tutorials_chinese(欢迎star)

Keras 是一个用于构建和训练深度学习模型的高阶 API。它可用于快速设计原型、高级研究和生产。 keras的3个优点：
方便用户使用、模块化和可组合、易于扩展

1.导入tf.keras

tensorflow2推荐使用keras构建网络，常见的神经网络都包含在keras.layer中(最新的tf.keras的版本可能和keras不同)

import tensorflow as tf
from tensorflow.keras import layers
print(tf.__version__)
print(tf.keras.__version__)

   
   
   
   1
2
3
4

2.构建简单模型

2.1模型堆叠

最常见的模型类型是层的堆叠：tf.keras.Sequential 模型

model = tf.keras.Sequential()
model.add(layers.Dense(32, activation='relu'))
model.add(layers.Dense(32, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))

   
   
   
   1
2
3
4

2.2网络配置

tf.keras.layers中网络配置：

activation：设置层的激活函数。此参数由内置函数的名称指定，或指定为可调用对象。默认情况下，系统不会应用任何激活函数。

kernel_initializer 和 bias_initializer：创建层权重（核和偏差）的初始化方案。此参数是一个名称或可调用对象，默认为 “Glorot uniform” 初始化器。

kernel_regularizer 和 bias_regularizer：应用层权重（核和偏差）的正则化方案，例如 L1 或 L2 正则化。默认情况下，系统不会应用正则化函数。

layers.Dense(32, activation='sigmoid')
layers.Dense(32, activation=tf.sigmoid)
layers.Dense(32, kernel_initializer='orthogonal')
layers.Dense(32, kernel_initializer=tf.keras.initializers.glorot_normal)
layers.Dense(32, kernel_regularizer=tf.keras.regularizers.l2(0.01))
layers.Dense(32, kernel_regularizer=tf.keras.regularizers.l1(0.01))

   
   
   
   1
2
3
4
5
6

3.训练和评估

3.1设置训练流程

构建好模型后，通过调用 compile 方法配置该模型的学习流程：

model = tf.keras.Sequential()
model.add(layers.Dense(32, activation='relu'))
model.add(layers.Dense(32, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))
model.compile(optimizer=tf.keras.optimizers.Adam(0.001),
             loss=tf.keras.losses.categorical_crossentropy,
             metrics=[tf.keras.metrics.categorical_accuracy])

   
   
   
   1
2
3
4
5
6
7

3.2输入Numpy数据

import numpy as np

train_x = np.random.random((1000, 72))
train_y = np.random.random((1000, 10))

val_x = np.random.random((200, 72))
val_y = np.random.random((200, 10))

model.fit(train_x, train_y, epochs=10, batch_size=100,
validation_data=(val_x, val_y))

3.3tf.data输入数据

dataset = tf.data.Dataset.from_tensor_slices((train_x, train_y))
dataset = dataset.batch(32)
dataset = dataset.repeat()
val_dataset = tf.data.Dataset.from_tensor_slices((val_x, val_y))
val_dataset = val_dataset.batch(32)
val_dataset = val_dataset.repeat()

model.fit(dataset, epochs=10, steps_per_epoch=30,
validation_data=val_dataset, validation_steps=3)

3.4评估与预测

test_x = np.random.random((1000, 72))
test_y = np.random.random((1000, 10))
model.evaluate(test_x, test_y, batch_size=32)
test_data = tf.data.Dataset.from_tensor_slices((test_x, test_y))
test_data = test_data.batch(32).repeat()
model.evaluate(test_data, steps=30)

 
 
 
 1
2
3
4
5
6

# predict
result = model.predict(test_x, batch_size=32)
print(result)

 
 
 
 1
2
3

4.构建高级模型

4.1函数式api

tf.keras.Sequential 模型是层的简单堆叠，无法表示任意模型。使用 Keras 函数式 API 可以构建复杂的模型拓扑，例如：

多输入模型，

多输出模型，

具有共享层的模型（同一层被调用多次），

具有非序列数据流的模型（例如，残差连接）。

使用函数式 API 构建的模型具有以下特征：

层实例可调用并返回张量。
输入张量和输出张量用于定义 tf.keras.Model 实例。
此模型的训练方式和 Sequential 模型一样。

input_x = tf.keras.Input(shape=(72,))
hidden1 = layers.Dense(32, activation='relu')(input_x)
hidden2 = layers.Dense(16, activation='relu')(hidden1)
pred = layers.Dense(10, activation='softmax')(hidden2)

model = tf.keras.Model(inputs=input_x, outputs=pred)
model.compile(optimizer=tf.keras.optimizers.Adam(0.001),
loss=tf.keras.losses.categorical_crossentropy,
metrics=[‘accuracy’])
model.fit(train_x, train_y, batch_size=32, epochs=5)

4.2模型子类化

通过对 tf.keras.Model 进行子类化并定义您自己的前向传播来构建完全可自定义的模型。在 init 方法中创建层并将它们设置为类实例的属性。在 call 方法中定义前向传播

class MyModel(tf.keras.Model):
    def __init__(self, num_classes=10):
        super(MyModel, self).__init__(name='my_model')
        self.num_classes = num_classes
        self.layer1 = layers.Dense(32, activation='relu')
        self.layer2 = layers.Dense(num_classes, activation='softmax')
    def call(self, inputs):
        h1 = self.layer1(inputs)
        out = self.layer2(h1)
        return out
<span class="token keyword">def</span> <span class="token function">compute_output_shape</span><span class="token punctuation">(</span>self<span class="token punctuation">,</span> input_shape<span class="token punctuation">)</span><span class="token punctuation">:</span>
    shape <span class="token operator">=</span> tf<span class="token punctuation">.</span>TensorShapej<span class="token punctuation">(</span>input_shape<span class="token punctuation">)</span><span class="token punctuation">.</span>as_list<span class="token punctuation">(</span><span class="token punctuation">)</span>
    shape<span class="token punctuation">[</span><span class="token operator">-</span><span class="token number">1</span><span class="token punctuation">]</span> <span class="token operator">=</span> self<span class="token punctuation">.</span>num_classes
    <span class="token keyword">return</span> tf<span class="token punctuation">.</span>TensorShape<span class="token punctuation">(</span>shape<span class="token punctuation">)</span>

model = MyModel(num_classes=10)
model.compile(optimizer=tf.keras.optimizers.RMSprop(0.001),
loss=tf.keras.losses.categorical_crossentropy,
metrics=[‘accuracy’])

model.fit(train_x, train_y, batch_size=16, epochs=5)

4.3自定义层

通过对 tf.keras.layers.Layer 进行子类化并实现以下方法来创建自定义层：

build：创建层的权重。使用 add_weight 方法添加权重。

call：定义前向传播。

compute_output_shape：指定在给定输入形状的情况下如何计算层的输出形状。
或者，可以通过实现 get_config 方法和 from_config 类方法序列化层。

class MyLayer(layers.Layer):
    def __init__(self, output_dim, **kwargs):
        self.output_dim = output_dim
        super(MyLayer, self).__init__(**kwargs)
<span class="token keyword">def</span> <span class="token function">build</span><span class="token punctuation">(</span>self<span class="token punctuation">,</span> input_shape<span class="token punctuation">)</span><span class="token punctuation">:</span>
    shape <span class="token operator">=</span> tf<span class="token punctuation">.</span>TensorShape<span class="token punctuation">(</span><span class="token punctuation">(</span>input_shape<span class="token punctuation">[</span><span class="token number">1</span><span class="token punctuation">]</span><span class="token punctuation">,</span> self<span class="token punctuation">.</span>output_dim<span class="token punctuation">)</span><span class="token punctuation">)</span>
    self<span class="token punctuation">.</span>kernel <span class="token operator">=</span> self<span class="token punctuation">.</span>add_weight<span class="token punctuation">(</span>name<span class="token operator">=</span><span class="token string">'kernel1'</span><span class="token punctuation">,</span> shape<span class="token operator">=</span>shape<span class="token punctuation">,</span>
                               initializer<span class="token operator">=</span><span class="token string">'uniform'</span><span class="token punctuation">,</span> trainable<span class="token operator">=</span><span class="token boolean">True</span><span class="token punctuation">)</span>
    <span class="token builtin">super</span><span class="token punctuation">(</span>MyLayer<span class="token punctuation">,</span> self<span class="token punctuation">)</span><span class="token punctuation">.</span>build<span class="token punctuation">(</span>input_shape<span class="token punctuation">)</span>

<span class="token keyword">def</span> <span class="token function">call</span><span class="token punctuation">(</span>self<span class="token punctuation">,</span> inputs<span class="token punctuation">)</span><span class="token punctuation">:</span>
    <span class="token keyword">return</span> tf<span class="token punctuation">.</span>matmul<span class="token punctuation">(</span>inputs<span class="token punctuation">,</span> self<span class="token punctuation">.</span>kernel<span class="token punctuation">)</span>

<span class="token keyword">def</span> <span class="token function">compute_output_shape</span><span class="token punctuation">(</span>self<span class="token punctuation">,</span> input_shape<span class="token punctuation">)</span><span class="token punctuation">:</span>
    shape <span class="token operator">=</span> tf<span class="token punctuation">.</span>TensorShape<span class="token punctuation">(</span>input_shape<span class="token punctuation">)</span><span class="token punctuation">.</span>as_list<span class="token punctuation">(</span><span class="token punctuation">)</span>
    shape<span class="token punctuation">[</span><span class="token operator">-</span><span class="token number">1</span><span class="token punctuation">]</span> <span class="token operator">=</span> self<span class="token punctuation">.</span>output_dim
    <span class="token keyword">return</span> tf<span class="token punctuation">.</span>TensorShape<span class="token punctuation">(</span>shape<span class="token punctuation">)</span>

<span class="token keyword">def</span> <span class="token function">get_config</span><span class="token punctuation">(</span>self<span class="token punctuation">)</span><span class="token punctuation">:</span>
    base_config <span class="token operator">=</span> <span class="token builtin">super</span><span class="token punctuation">(</span>MyLayer<span class="token punctuation">,</span> self<span class="token punctuation">)</span><span class="token punctuation">.</span>get_config<span class="token punctuation">(</span><span class="token punctuation">)</span>
    base_config<span class="token punctuation">[</span><span class="token string">'output_dim'</span><span class="token punctuation">]</span> <span class="token operator">=</span> self<span class="token punctuation">.</span>output_dim
    <span class="token keyword">return</span> base_config

@<span class="token builtin">classmethod</span>
<span class="token keyword">def</span> <span class="token function">from_config</span><span class="token punctuation">(</span>cls<span class="token punctuation">,</span> config<span class="token punctuation">)</span><span class="token punctuation">:</span>
    <span class="token keyword">return</span> cls<span class="token punctuation">(</span><span class="token operator">**</span>config<span class="token punctuation">)</span>

model = tf.keras.Sequential(
[
MyLayer(10),
layers.Activation(‘softmax’)
])

model.compile(optimizer=tf.keras.optimizers.RMSprop(0.001),
loss=tf.keras.losses.categorical_crossentropy,
metrics=[‘accuracy’])

model.fit(train_x, train_y, batch_size=16, epochs=5)

4.3回调

callbacks = [
    tf.keras.callbacks.EarlyStopping(patience=2, monitor='val_loss'),
    tf.keras.callbacks.TensorBoard(log_dir='./logs')
]
model.fit(train_x, train_y, batch_size=16, epochs=5,
         callbacks=callbacks, validation_data=(val_x, val_y))

 
 
 
 1
2
3
4
5
6

5保持和恢复

5.1权重保存

model = tf.keras.Sequential([
layers.Dense(64, activation='relu'),
layers.Dense(10, activation='softmax')])

model.compile(optimizer=tf.keras.optimizers.Adam(0.001),
loss=‘categorical_crossentropy’,
metrics=[‘accuracy’])

model.save_weights(’./weights/model’)
model.load_weights(’./weights/model’)
model.save_weights(’./model.h5’)
model.load_weights(’./model.h5’)

5.2保存网络结构

# 序列化成json
import json
import pprint
json_str = model.to_json()
pprint.pprint(json.loads(json_str))
fresh_model = tf.keras.models.model_from_json(json_str)

 
 
 
 1
2
3
4
5
6

# 保持为yaml格式  #需要提前安装pyyaml

yaml_str = model.to_yaml()
print(yaml_str)
fresh_model = tf.keras.models.model_from_yaml(yaml_str)

5.3保存整个模型

model = tf.keras.Sequential([
  layers.Dense(10, activation='softmax', input_shape=(72,)),
  layers.Dense(10, activation='softmax')
])
model.compile(optimizer='rmsprop',
              loss='categorical_crossentropy',
              metrics=['accuracy'])
model.fit(train_x, train_y, batch_size=32, epochs=5)
model.save('all_model.h5')
model = tf.keras.models.load_model('all_model.h5')

 
 
 
 1
2
3
4
5
6
7
8
9
10

6.将keras用于Estimator

Estimator API 用于针对分布式环境训练模型。它适用于一些行业使用场景，例如用大型数据集进行分布式训练并导出模型以用于生产

model = tf.keras.Sequential([layers.Dense(10,activation='softmax'),
                          layers.Dense(10,activation='softmax')])

model.compile(optimizer=tf.keras.optimizers.RMSprop(0.001),
loss=‘categorical_crossentropy’,
metrics=[‘accuracy’])

estimator = tf.keras.estimator.model_to_estimator(model)

                                </div>
            <link href="https://csdnimg.cn/release/phoenix/mdeditor/markdown_views-e44c3c0e64.css" rel="stylesheet">
                </div>