0
$\begingroup$

I desperately need help regarding ML/NN models that would be appropriate for binary input data..

So, I have an image dataset in which [R,G,B] values can only take binary values (that is 0 and 1). For example, a pixel value can be [0, 1, 1], or [1, 0, 0], or [0, 0, 0], or [.., .., ..] etc. Essentially, this makes an image to consist of only of the following colors : true red, true green, true blue, true grey, true yellow, true magenta, true black, and true white. What is more, I would like to perform binary classification for this dataset.

For example, this is an array of an image:

[[[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  ...
  ...
  [1. 0. 0.]
  [1. 0. 0.]
  [1. 0. 0.]]

 [[1. 1. 1.]
  [1. 1. 1.]
  [1. 1. 1.]
  ...
  [1. 0. 0.]
  [1. 0. 0.]
  [1. 0. 0.]]

  ...
  [0. 0. 0.]
  [0. 0. 0.]
  [0. 0. 0.]]] 

Which ML (say SVM) or DL model would be appropriate for such a task? I am not sure CNN would be an appropriate approach, even if I am dealing with images, as I doubt NN can work with binary input data. In fact, I performed a simple NN (not CNN) on them and all y labels were predicted to be 0. However, I am not sure if my implementation was correct because I don't have much experience in the field (and maybe my NN was wrong). Please correct me if I am wrong. Does anybody have an idea of how I could make this could work?

I forgot to mention that these are 256x256 images and I think the end dataset will be quite large (>5000), in case that info is needed.

Please help!! Thank you so much in advance!

$\endgroup$
1

1 Answer 1

1
$\begingroup$

Amadeo Amadei. In my opinion, if you have labelled training data, the CNN will be good approach. CNN works well on the binary image as well as other RGB images. If you don't have labelled data, you can try Unsupervised Learning algorithms.

$\endgroup$
1
  • $\begingroup$ Thank you for your comment! $\endgroup$ Sep 22, 2022 at 8:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.