confusion matrix doesn't match about dd-net

[fandulu]

Hi, I am not sure the setting on your site, maybe it is related to issue 5, could be due to the insufficient training.

[chillcloud-dev]

Hello @fandulu

Thanks for your fast reply.

I thought it was something wrong with the settings since I just tried the provided code with the default values. And even changing the values I didn't get the desired confusion matrix.

Strangely I get similar model accuracy like in issue 5 , however the confusion matrix looks like this

having the following values:

[[54 1 0 3 0 0 0 0 0 0 0 0 0 0]
[ 4 47 0 2 1 1 0 0 0 5 0 0 0 1]
[ 0 2 49 0 0 0 0 0 4 0 0 0 0 0]
[ 6 0 0 43 0 2 0 0 0 0 0 0 0 0]
[ 0 0 0 0 54 0 1 0 0 0 0 0 0 0]
[ 0 0 0 0 0 57 0 1 0 0 0 0 0 0]
[ 0 0 0 0 0 0 62 0 0 0 0 0 0 0]
[ 0 0 0 0 1 0 0 53 0 0 0 0 0 0]
[ 0 2 1 0 0 0 0 0 64 1 0 0 0 0]
[ 0 2 0 3 0 0 0 0 0 56 0 0 0 0]
[ 0 0 0 0 0 1 0 0 0 0 68 0 0 0]
[ 0 0 0 0 0 0 0 0 0 0 0 57 0 0]
[ 0 0 0 0 0 0 0 0 0 0 0 0 58 0]
[ 0 1 1 0 2 0 0 0 0 0 0 0 0 69]]

More training didn't change the values much. This was the best I could get.
What I am expecting is something similar to :

So my question was, how you get this confusion matrix, where you achieve values >90% ?
I am using the provided code, so I expected to have a similar confusion matrix values.

Thanks for

[fandulu]

I think you need to normalize the value to be a ratio of 0%-100% instead of the number of samples.

[chillcloud-dev]

okay
if i normalize them i reach the values from the paper

@fandulu thank you so much