vdumoulin / conv_arithmetic

A technical report on convolution arithmetic in the context of deep learning

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Error in convolution computation

ludovic-carre opened this issue · comments

Aren't there mistakes in the images 1 to 8 for numerical_padding_strides-*.pdf ? For instance shouldn't the second element of the first row be 5 ?
Here is what is computed:

                   0 0 0 0 0 0 0                                               1 6 5
                   0 2 2 3 3 3 0                    Original result:           7 10 9
2 0 1              0 0 1 3 0 3 0                                               7 10 8
1 0 0     *        0 2 3 0 1 3 0       ---->
0 1 1              0 3 3 2 1 2 0                                               1 5 6
                   0 3 3 0 2 3 0                    My result:                 7 8 3
                   0 0 0 0 0 0 0                                               3 10 4

If I am correct then the mistakes are also on this website: http://deeplearning.net/software/theano/tutorial/conv_arithmetic.html

PS: Sorry for the chaotic display, can't figure it out.

What is your stride for your example?

Stride 2 as you can see on the website linked.

Your result is correct.

Hi @PiggyGenius , thank you for your bug report.
I took the liberty of editing your comment to fix the display :)

I can confirm that you computed the correct outcome for a stride 2 convolution and that the second animation of the Theano tutorial is indeed incorrect. We shall fix that. (@vdumoulin I don't think I have access anymore. Do you? Drop me a line when you have time.)

By 1 to 8 for numerical_padding_strides-*.pdf do you mean the animations generated by the code? I didn't have time to re-run the code myself, but we will look into it as soon as we have some time.

Thank you again for the bug report.

I was referring to the images in the pdf folder of this repository but I apparently made a mistakes. I thought they were the same images as the ones of the theano gif.

Great, thanks for reporting this. We will look into it as soon as possible.