every blog every motto: Whatever is worth doing is worth doing well.
0. 前言
实战神经网络,selu激活函数,自带归一化。
1. 代码部分
1. 导入模块
import matplotlib as mpl
import matplotlib.pyplot as plt
%matplotlib inline
import numpy as np
import sklearn
import pandas as pd
import os
import sys
import time
import tensorflow as tf
from tensorflow import keras
print(tf.__version__)
print(sys.version_info)
for module in mpl,np,pd,sklearn,tf,keras:
print(module.__name__,module.__version__)
2. 读取数据
fashion_mnist = keras.datasets.fashion_mnist
# print(fashion_mnist)
(x_train_all,y_train_all),(x_test,y_test) = fashion_mnist.load_data()
x_valid,x_train = x_train_all[:5000],x_train_all[5000:]
y_valid,y_train = y_train_all[:5000],y_train_all[5000:]
# 打印格式
print(x_valid.shape,y_valid.shape)
print(x_train.shape,y_train.shape)
print(x_test.shape,y_test.shape)
3. 数据归一化
# 数据归一化
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
# x_train:[None,28,28] -> [None,784]
x_train_scaled = scaler.fit_transform(x_train.astype(np.float32).reshape(-1,1)).reshape(-1,28,28)
x_valid_scaled = scaler.transform(x_valid.astype(np.float32).reshape(-1,1)).reshape(-1,28,28)
x_test_scaled = scaler.transform(x_test.astype(np.float32).reshape(-1,1)).reshape(-1,28,28)
4. 构建模型
注意:激活函数selu自带归一化功能
# tf.keras.models.Sequential()
# 构建模型
# 深度神经网络
model = keras.models.Sequential()
# 输入数据展平
model.add(keras.layers.Flatten(input_shape=[28,28]))
# 隐藏层 20层
for _ in range(20):
model.add(keras.layers.Dense(100,activation="selu")) # 激活函数自带归一化
# 输出层
model.add(keras.layers.Dense(10,activation="softmax"))
#
model.compile(loss='sparse_categorical_crossentropy',optimizer='sgd',metrics=['accuracy'])
model.summary()
5. 训练
# 回调函数 Tensorboard(文件夹)\earylystopping\ModelCheckpoint(文件名)
logdir = os.path.join("dnn-selu-callbacks")
print(logdir)
if not os.path.exists(logdir):
os.mkdir(logdir)
# 文件名
output_model_file = os.path.join(logdir,"fashion_mnist_model.h5")
callbacks = [
keras.callbacks.TensorBoard(logdir),
keras.callbacks.ModelCheckpoint(output_model_file,save_best_only=True),
keras.callbacks.EarlyStopping(patience=5,min_delta=1e-3),
]
# 开始训练
history = model.fit(x_train_scaled,y_train,epochs=10,validation_data=(x_valid_scaled,y_valid),callbacks=callbacks)
6. 学习曲线
# 画图
def plot_learning_curves(history):
pd.DataFrame(history.history).plot(figsize=(8,5))
plt.grid(True)
plt.gca().set_ylim(0,1)
plt.show()
plot_learning_curves(history)
# 损失函数,刚开始下降慢的原因
# 1. 参数众多,训练不充分
# 2. 梯度消失 -》 链式法则中
# 解决: selu缓解梯度消失
7. 测试集上
model.evaluate(x_test_scaled,y_test)
8. 问题
selu(激活函数)缓解梯度消失问题