Implementing backward-convolution in CNN for multi-channel data

Question

I have been trying to get a deeper understanding of convolutional operation as I am implementing a convolutional neural network. But I am stuck while trying to calculate the backward pass or deconvolution.

Lets say the input is a 3 dimensional RGB image with dimension 3x7x7 The filter has the dimension 3x3x3. On convolving with stride set to 2 we will get an output of dimension 3x3.

Now here is my problem. I have read that deconvolution is the convolution of the output with flipped kernel. But on flipping the kernel, it will still be of dimension 3x3x3 and output is of dimension 3x3 which . The input was of dimension 3x7x7 . So, how is deconvolution calculated ?

Are you talking about 2D or 3D convolution? What do you mean by 3x3x3 filter shape? — Maxim
@maxim I am talking about 3D convolution thats why the filter is 3D — Ayush Chaurasia
@Maxim I believe he means 2D color image, the 3 is for the 3 channels — Vj-

Lukasz Tracewski Lukasz Tracewski · Accepted Answer · 2017-10-13T07:03:04

Here is a nice visualisation how convolution and deconvolution (transposed convolution). The white pieces are simply zeros.

Implementing backward-convolution in CNN for multi-channel data

1 Answers