Python Keras 中的 RMSE/RMSLE 损失函数
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/43855162/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
RMSE/ RMSLE loss function in Keras
提问by dennis
I try to participate in my first Kaggle competition where RMSLEis given as the required loss function. For I have found nothing how to implement this loss functionI tried to settle for RMSE. I know this was part of Kerasin the past, is there any way to use it in the latest version, maybe with a customized function via backend?
我尝试参加我的第一次 Kaggle 比赛，其中RMSLE给出了所需的损失函数。因为我没有发现如何实现这一点，loss function我试图解决这个问题RMSE。我知道这是Keras过去的一部分，有没有办法在最新版本中使用它，也许通过自定义功能backend？
This is the NN I designed:
这是我设计的神经网络：
from keras.models import Sequential
from keras.layers.core import Dense , Dropout
from keras import regularizers
model = Sequential()
model.add(Dense(units = 128, kernel_initializer = "uniform", activation = "relu", input_dim = 28,activity_regularizer = regularizers.l2(0.01)))
model.add(Dropout(rate = 0.2))
model.add(Dense(units = 128, kernel_initializer = "uniform", activation = "relu"))
model.add(Dropout(rate = 0.2))
model.add(Dense(units = 1, kernel_initializer = "uniform", activation = "relu"))
model.compile(optimizer = "rmsprop", loss = "root_mean_squared_error")#, metrics =["accuracy"])
model.fit(train_set, label_log, batch_size = 32, epochs = 50, validation_split = 0.15)
I tried a customized root_mean_squared_errorfunction I found on GitHub but for all I know the syntax is not what is required. I think the y_trueand the y_predwould have to be defined before passed to the return but I have no idea how exactly, I just started with programming in python and I am really not that good in math...
我尝试了root_mean_squared_error在 GitHub 上找到的自定义函数，但我知道语法不是必需的。我认为 they_true和 they_pred必须在传递给 return 之前定义，但我不知道具体如何，我刚开始用 python 编程，我真的不太擅长数学......
from keras import backend as K
def root_mean_squared_error(y_true, y_pred):
return K.sqrt(K.mean(K.square(y_pred - y_true), axis=-1))
I receive the following error with this function:
我收到此功能的以下错误：
ValueError: ('Unknown loss function', ':root_mean_squared_error')
Thanks for your ideas, I appreciate every help!
感谢您的想法，我感谢每一个帮助！
回答by Dr. Snoopy
When you use a custom loss, you need to put it without quotes, as you pass the function object, not a string:
当您使用自定义损失时，您需要将其不带引号，因为您传递的是函数对象，而不是字符串：
def root_mean_squared_error(y_true, y_pred):
return K.sqrt(K.mean(K.square(y_pred - y_true)))
model.compile(optimizer = "rmsprop", loss = root_mean_squared_error,
metrics =["accuracy"])
回答by Germán Sanchis
The accepted answer contains an error, which leads to that RMSE being actually MAE, as per the following issue:
接受的答案包含一个错误，导致 RMSE 实际上是 MAE，根据以下问题：
https://github.com/keras-team/keras/issues/10706
https://github.com/keras-team/keras/issues/10706
The correct definition should be
正确的定义应该是
def root_mean_squared_error(y_true, y_pred):
return K.sqrt(K.mean(K.square(y_pred - y_true)))
回答by Richard Xue
If you are using latest tensorflow nightly, although there is no RMSE in the documentation, there is a tf.keras.metrics.RootMeanSquaredError()in the source code.
如果你每晚使用最新的 tensorflow，虽然文档中没有 RMSE，但tf.keras.metrics.RootMeanSquaredError()在源代码中有一个。
sample usage:
示例用法：
model.compile(tf.compat.v1.train.GradientDescentOptimizer(learning_rate),
loss=tf.keras.metrics.mean_squared_error,
metrics=[tf.keras.metrics.RootMeanSquaredError(name='rmse')])
回答by George C
I prefer reusing part of the Keras work
我更喜欢重用部分 Keras 工作
from keras.losses import mean_squared_error
def root_mean_squared_error(y_true, y_pred):
return K.sqrt(mean_squared_error(y_true, y_pred))
model.compile(optimizer = "rmsprop", loss = root_mean_squared_error,
metrics =["accuracy"])
