I am fairly new to machine learning. I am trying to generate a new image from other images of the same shape. An example of the image I'm trying to generate is an Hi-C data matrix: https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcT5_h4AXnOD3HReBIjZA8vBuv-DE-kedSVA4Q&usqp=CAU. I have many passes of an image and I am trying to recover the initial image from such passes (which would in turn have the same resolution).

I am aware of GANs but I could not find an example of them being used with multiple images as inputs instead of only one. I thought of using a normal network and feeding along with each pixel, the value of the neighboring ones but that would mean having thousands of variables as inputs. This is because, in a normal scenario I would have 14 images and, for each pixel I could have anywhere from (8 to 120) * 14 additional inputs (assuming a radius of 3 to 11).

And, at the same time I would like to include information about each neighbouring pixel as for the problem I am working on, a single data point is influenced by neighbouring ones.

Is what I am trying to achieve possible? Is there any examples of such network architecture being used somewhere?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.