Felix

Felix | 27 | MSc in Radiohysics | Data Architect

Distributed Systems + Machine Learning + Forecasting + Big Data Infrastructure

[Felix @felix_hsde]

@DonetskIsoir I was told that this person is neither. They are apparently just some mentally ill American who likes attention.

[Felix @felix_hsde]

@DigitalHastur @RealAidenCook True GNU way

[Felix @felix_hsde]

@web_of_throats I agree with more body types.
I do disagree with sliders though - too much work and might need like 2 years after release to fully fix all possible bugs that could occur. WoW engine doesn't seem to be able to handle morphing with character models.

[Felix @felix_hsde]

@ZyMazza Not sure how those are connected. Planck length being some "spacetime pixel-length" is a misconception.

[Felix @felix_hsde]

@Cantor_Paradise Encountered fractional derivatives only during complex analysis as a part of the paragraph going over Cauchy's integral formula for derivatives.
Never seen it otherwise.

[Felix @felix_hsde]

@TaliaRinger That's just Bayesian reasoning. The underlying problem with the specified probabilities there is also something that is intrinsic to it.

[Felix @felix_hsde]

@WKCosmo You should avoid developing a mindset that "laws of deductive logic and their consequences are properties of our universe" though. In the way you've stated it - I agree. But making it stronger can lead to borderline theological thinking.

[グラフ理論 @graphtheory]

The reason most people look for an ulterior motive for <argument> is they don’t agree with <conclusion>. But if you try assuming <conclusion> then <argument> makes sense.

[Felix @felix_hsde]

@norpadon @teortaxesTex You are 100% using a whole tabular NNs on 20mil rows dataset with like maybe 10-15 features for simple B2C churn.
I trust your expertise.

[Felix @felix_hsde]

IDK why people worship academia personalities.
Like they are not some supermen who are "holier than thou" or superior to others. They are still mostly the same people who value their own asses above everything else and can say stupid shit too.
And mathematicians are not safe from that - I've seen so many French and German math profs who are well established in their research fields and then also believe is stuff like Ether theory conspiracy (aka global STO cabal that hides the truth about ether dynamics) or in fringe schizoid historical conspiracies.

[Felix @felix_hsde]

@norpadon @teortaxesTex @predict_addict

[Felix @felix_hsde]

@the_posterior @teortaxesTex Good. Just don't skip the underlying theory.
Too many people who ignore it and then just struggle with the prod data.
Highly recommend also looking into conformal prediction.

[Felix @felix_hsde]

@norpadon @teortaxesTex Guess you enjoy overfitting and don't really care about bounds on capacity.

[Felix @felix_hsde]

@the_posterior @teortaxesTex Statistical learning theory. I worked with ML before the hype back in 2018 in CERN-affiliated lab.
They kinda expected you to know formal theory behind ML

[Felix @felix_hsde]

@norpadon @teortaxesTex ...and then you end up with people who just throw XGBoost or NNs at everything and end up with like ~50% accuracy and bad F1 scores with production data after just 1 month.
If you are unwilling to understand why most of those algorithms work - kudos to you.

[Felix @felix_hsde]

@perrymetzger That's what happens when you are stupid but trying to sound smart: just write a lot and throw as many buzzwords and technical terms as possible.

[Felix @felix_hsde]

@noinconsistency Dates and timestamps are a constant problem.
Like try migrating from one DB to another and you will get it.

[Felix @felix_hsde]

@relionwastaken Same people who used to dress up as elves or orcs and fight with plastic swords in parks.
What do you expect.