Neural Network Classification Results

Hi Guys,

Im training a NN using the matlab toolbox and I have selected the data partitioning to random data division.(i.e 70:15:15,Train:Val:Test).

When I look at the confusion matrix at the end of the process(attached), I notice its used more data for the classification process for one class compared to the other.

Can anyone advise why this could be happening?

Answer 1

[awezmm]

Are you doing semantic segmentation or image classification?

The confusion matrix helps indicate differences in classifications between true positives, true negatives, false positives, false negatives. It does NOT indicate how much total data was used in training. Rather, it can show how well classification of specific classes is performing.

The details on the large plot in this page gives a better explanation: https://www.mathworks.com/help/deeplearning/ref/plotconfusion.html

Answer 2

[Greg Heath]

The original class sizes are unequal.

Hope this helps

THANK YOU FOR FORMALLY ACEEPTING MY ANSWER

Greg