Debraj Basu

@and_nini @ValerioGherardi @shunishiha I am reading this work and found it very interesting, is there an implementation already available for this?

I'm extremely excited to announce the pre-print of our new paper: "Authorship Verification based on the Likelihood Ratio of Grammar Models" arxiv.org/abs/2403.08462… with Oren Halvani, Lukas Graner, @ValerioGherardi and @shunishiha

@ylecun @kareem_carr If we can observe "all" variables, then the ground truth label is often indeed a deterministic result.

@ylecun @kareem_carr In the traditional statistical learning setting where the goal is to find the Bayes optimal classifier, we refer to the underlying distribution which in most cases is an artefact of the unavailability of access to the unknown/unseen variables and not because of "true randomness".

Help! I’m living a lie. I talk about flipping coins. I say the probability of heads is 0.5 but I’ve never actually flipped a coin. I bluffed my way into a Harvard stats PhD and I still have no clue what happens when you flip a coin...and at this point, I’m too scared to find out.