首先保存
1、模型结构,
2、模型参数数据
#!/usr/bin/env python
# -*- coding:utf-8 -*-
# Author: Jia ShiLin
'''
#模型的权重保存在HDF5中
#模型的结构保存在JSON文件或者YAML文件中
'''
'''
#Keras提供了to_json()生成模型JSON描述,并将模型的JSON描述保存到文件中,
#反序列化时候,通过model_from_json()函数加载模型描述,编译生成模型
##save_weights()函数可以保存模型的权重值,加载时使用load_weights()
##eg:训练一个模型,使用json格式描述模型结构,保存model.json文件中,权重信息保存到本地目录model.json.h5文件中
###当新的数据需要预测时候,加载模型和权重信息,创建新模型,必须先编译,
'''
from sklearn import datasets
import numpy as np
from keras.models import Sequential
from keras.layers import Dense
from keras.utils import to_categorical
from keras.models import model_from_json
dataset = datasets.load_iris()
x = dataset.data
y = dataset.target
y_labels = to_categorical(y, num_classes=3)
seed = 7
np.random.seed(seed)
# 构建模型函数
def create_model(optimizer='rmsprop', init='glorot_uniform'):
model = Sequential()
model.add(Dense(units=4, activation='relu', input_dim=4, kernel_initializer=init))
model.add(Dense(units=6, activation='relu', kernel_initializer=init))
model.add(Dense(units=3, activation='relu', kernel_initializer=init))
# 编译模型,
model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=['accuracy'])
return model
# 构建模型
model = create_model()
model.fit(x, y_labels, epochs=200, batch_size=50, verbose=0)
scores = model.evaluate(x, y_labels, verbose=0)
print(model.metrics_names[1], scores[1] * 100)
# 保存模型
model_json = model.to_json()
with open('./data/model.json', 'w') as file:
file.write(model_json)
# 保存模型权重值
model.save_weights('./data/model.json.h5')
接着读取模型,和参数
#!/usr/bin/env python
# -*- coding:utf-8 -*-
# Author: Jia ShiLin
from sklearn import datasets
import numpy as np
from keras.models import Sequential
from keras.layers import Dense
from keras.utils import to_categorical
from keras.models import model_from_json
# 从json文件中加载模型
with open('./data/model.json', 'r') as file:
model_json = file.read()
new_model = model_from_json(model_json)
new_model.load_weights('./data/model.json.h5')
#编译模型!!!!
#new_model.compile(loss = 'categorical_crossentropy',optimizer='rmsprop',metrics=['accuracy'])
new_model.compile(loss = 'mse',optimizer='rmsprop',metrics=['accuracy'])
dataset = datasets.load_iris()
x = dataset.data
y = dataset.target
y_labels = to_categorical(y, num_classes=3)
scores =new_model.evaluate(x,y_labels,verbose=0)
print(new_model.metrics_names[1], scores[1] * 100)
原文链接:https://blog.csdn.net/qq_35290785/article/details/95245155