• 10 Posts
  • 952 Comments
Joined 1 year ago
cake
Cake day: June 6th, 2023

help-circle

  • (nice ad hominem) Christ. When you reduce a high dimensional object into an embedded space, yes you keep only the first N features, but those N features are the most variable, and the loadings they contain can be used to map back to (a very good) approximation of the source images. It’s akin to reverse engineering a very lossy compression to something that (very strongly) resembles the source image (otherwise feature extraction wouldn’t be useful), and it’s entirely doable.

















  • Thanks for the Anton Petrov recommendation. I’ve been looking for better up to date science reviews, and I was getting sick of the Two Minute Papers channel with its near-constant shilling for Nvidia and OpenAI.

    For me, a daily thing I do is sit in my grandmothers armchair, and put my feet up on a small table that also belonged to her, with my laptop on my lap. I realise that sitting is not a super exciting task, but I guess sitting draped in the unconditional love of your deceased grandma’s possessions makes it sound a bit better.