October 2018
Intermediate to advanced
472 pages
10h 57m
English
GANs are just downright cool. When we look back at the skills and intuition we've built throughout these projects, we had an interesting idea. Could we predict missing information? Or, stated in a different way: can we create data that should be in an image, but that's not there? If we can take text input and generate novel text output, and if we can take a 2D image and generate or predict a 3D positional output, then it would seem possible that, if we have a 2D image that's missing some information, maybe we ought to be able to generate the missing information? So, in this chapter, we built a neural network that fills in the missing part of a handwritten digit. We previously built ...
Read now
Unlock full access