[Bromskloss]

Those images [0] inputs that were optimised to maximise a certain classification response were cool! Instead of going to this peak of the response function, is there a way to explore the shell where the actual images reside? Would such images look, to our eyes, more like real input than the optimised input? I suspect they won't, but I still would like to see what is between the dogs!

[0] http://www.inference.vc/content/images/2017/11/Screen-Shot-2...

[lebek]

This paper is sort of about that: https://arxiv.org/abs/1710.11381

They use a gamma distribution which has more probability density near the origin, which causes samples around the origin and interpolations to be more like real input.

[tobr]

The source is a great read: https://distill.pub/2017/feature-visualization/