Keras 中的自定义加权损失函数，用于对每个元素进行加权

本文介绍了Keras 中的自定义加权损失函数，用于对每个元素进行加权的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！

问题描述

我正在尝试创建一个简单的加权损失函数.

I'm trying to create a simple weighted loss function.

比如说，我的输入维度是 100 * 5，输出维度也是 100 * 5.我还有一个相同维度的权重矩阵.

Say, I have input dimensions 100 * 5, and output dimensions also 100 * 5. I also have a weight matrix of the same dimension.

类似于以下内容:

import numpy as np
train_X = np.random.randn(100, 5)
train_Y = np.random.randn(100, 5)*0.01 + train_X

weights = np.random.randn(*train_X.shape)

定义自定义损失函数

def custom_loss_1(y_true, y_pred):
    return K.mean(K.abs(y_true-y_pred)*weights)

定义模型

from keras.layers import Dense, Input
from keras import Model
import keras.backend as K

input_layer = Input(shape=(5,))
out = Dense(5)(input_layer)
model = Model(input_layer, out)

使用现有指标进行测试工作正常

model.compile('adam','mean_absolute_error')
model.fit(train_X, train_Y, epochs=1)

使用我们的自定义损失函数进行测试不起作用

model.compile('adam',custom_loss_1)
model.fit(train_X, train_Y, epochs=10)

它提供了以下堆栈跟踪:

It gives the following stack trace:

InvalidArgumentError (see above for traceback): Incompatible shapes: [32,5] vs. [100,5]
 [[Node: loss_9/dense_8_loss/mul = Mul[T=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:CPU:0"](loss_9/dense_8_loss/Abs, loss_9/dense_8_loss/mul/y)]]

数字 32 是从哪里来的?

Where is the number 32 coming from?

def custom_loss_2(y_true, y_pred):
    return K.mean(K.abs(y_true-y_pred)*K.ones_like(y_true))

这个函数似乎可以完成工作.因此，可能表明将 Keras 张量作为权重矩阵会起作用.所以，我创建了另一个版本的损失函数.

This function seems to do the work. So, probably suggests that a Keras tensor as a weight matrix would work. So, I created another version of the loss function.

from functools import partial

def custom_loss_3(y_true, y_pred, weights):
    return K.mean(K.abs(y_true-y_pred)*K.variable(weights, dtype=y_true.dtype))

cl3 = partial(custom_loss_3, weights=weights)

使用 cl3 拟合数据会产生与上述相同的错误.

Fitting data using cl3 gives the same error as above.

InvalidArgumentError (see above for traceback): Incompatible shapes: [32,5] vs. [100,5]
     [[Node: loss_11/dense_8_loss/mul = Mul[T=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:CPU:0"](loss_11/dense_8_loss/Abs, loss_11/dense_8_loss/Variable/read)]]

我想知道我错过了什么！我本可以在 Keras 中使用 sample_weight 的概念；但随后我必须将我的输入重塑为 3d 矢量.

I wonder what I'm missing! I could have used the notion of sample_weight in Keras; but then I'd have to reshape my inputs to a 3d vector.

我认为这个自定义损失函数应该很简单.

I thought that this custom loss function should really have been trivial.

推荐答案

在 model.fit 中，batch size 默认为 32，这就是这个数字的来源.这是发生的事情:

In model.fit the batch size is 32 by default, that's where this number is coming from. Here's what's happening:

在 custom_loss_1 中，张量 K.abs(y_true-y_pred) 的形状为 (batch_size=32, 5)，而numpy 数组 weights 的形状为 (100, 5).这是无效的乘法，因为维度不一致且无法应用广播.

In custom_loss_1 the tensor K.abs(y_true-y_pred) has shape (batch_size=32, 5), while the numpy array weights has shape (100, 5). This is an invalid multiplication, since the dimensions don't agree and broadcasting can't be applied.

在 custom_loss_2 中不存在此问题，因为您将 2 个具有相同形状的张量相乘 (batch_size=32, 5).

In custom_loss_2 this problem doesn't exist because you're multiplying 2 tensors with the same shape (batch_size=32, 5).

在 custom_loss_3 中，问题与在 custom_loss_1 中相同，因为将 weights 转换为 Keras 变量不会改变它们的形状.

In custom_loss_3 the problem is the same as in custom_loss_1, because converting weights into a Keras variable doesn't change their shape.

更新:您似乎想为每个训练样本中的每个元素赋予不同的权重，因此 weights 数组应具有形状 (100, 5) 确实如此.在这种情况下，我会将您的权重数组输入您的模型，然后在损失函数中使用此张量:

UPDATE: It seems you want to give a different weight to each element in each training sample, so the weights array should have shape (100, 5) indeed.In this case, I would input your weights' array into your model and then use this tensor within the loss function:

import numpy as np
from keras.layers import Dense, Input
from keras import Model
import keras.backend as K
from functools import partial


def custom_loss_4(y_true, y_pred, weights):
    return K.mean(K.abs(y_true - y_pred) * weights)


train_X = np.random.randn(100, 5)
train_Y = np.random.randn(100, 5) * 0.01 + train_X
weights = np.random.randn(*train_X.shape)

input_layer = Input(shape=(5,))
weights_tensor = Input(shape=(5,))
out = Dense(5)(input_layer)
cl4 = partial(custom_loss_4, weights=weights_tensor)
model = Model([input_layer, weights_tensor], out)
model.compile('adam', cl4)
model.fit(x=[train_X, weights], y=train_Y, epochs=10)

这篇关于Keras 中的自定义加权损失函数，用于对每个元素进行加权的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！