I would not say it’s impossible… my lab is working on this (https://arxiv.org/abs/2405.14577 https://arxiv.org/abs/2405.14577) and though it’s far from mature -

I would not say it’s impossible… my lab is working on this (https://arxiv.org/abs/2405.14577 https://arxiv.org/abs/2405.14577) and though it’s far from mature - in theory some kind of resistance to downstream training isn’t impossible. I think under classical statistical learning theory you would predict it’s impossible with unlimited training data and budget for searching for models but we don’t have those same gaurentees with deep neural networks.