Machine Learning Idea – Add And Remove Noise Points By Different Distribution Functions

The idea is to see if there exist an idea to shape noise. Not only by adding points with random value but also to remove points according some other distribution.

The remove part is not as problematic as I think. What if you have a filled 2D plane and remove points according to the random function. First maybe you need to add some points. Then do some removal.

So instead of np.random.rand(100) or something like that I’m thinking a 2D image. With some adapted width and height as parameters to capture a noise image.

Hmm. Could be interesting to see if ?negative noise could be used in a generator(neg. Noise) GAN network.