TensorFlow的`conv2d_transpose（）`操作有什么作用？ [英] What does TensorFlow's `conv2d_transpose()` operation do?

问题描述

conv2d_transpose（）操作的文档没有明确说明其作用：

The documentation for the conv2d_transpose() operation does not clearly explain what it does:

conv2d的转置。

The transpose of conv2d.

此操作有时在
反卷积网络，但实际上是
conv2d的转置（梯度），而不是实际的反卷积。

This operation is sometimes called "deconvolution" after Deconvolutional Networks, but is actually the transpose (gradient) of conv2d rather than an actual deconvolution.

我浏览了文档指向的文件，但没有帮助。

I went through the paper that the doc points to, but it did not help.

是什么

推荐答案

这是我所见过的最好的解释在线卷积转置的工作原理是此处。

This is the best explanation I've seen online how convolution transpose works is here.

我将给出自己的简短描述。它采用分数步长进行卷积。换句话说，将输入值分隔为零（以零表示），以将过滤器应用于可能小于过滤器尺寸的区域。

I'll give my own short description. It applies convolution with a fractional stride. In other words spacing out the input values (with zeroes) to apply the filter over a region that's potentially smaller than the filter size.

为什么要使用它。与双线性插值法或某种其他固定形式的上采样相反，它可以用作一种具有学习权重的上采样。

As for the why one would want to use it. It can be used as a sort of upsampling with learned weights as opposed to bilinear interpolation or some other fixed form of upsampling.

这篇关于TensorFlow的`conv2d_transpose（）`操作有什么作用？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IT屋！