iiinova — where your digital self keeps evolving

Reposted by Richi Paul

Paper Skygest Team @paper-feed.bsky.social · 19/08/2025

**Please repost** If you're enjoying Paper Skygest -- our personalized feed of academic content on Bluesky -- we'd appreciate you reposting this! We’ve found that the most effective way for us to reach new users and communities is through users sharing it with their network

2114143

Richi Paul @erpel.bsky.social · 05/05/2026

seems like you can actually change it

030

Reposted by Richi Paul

Symposium on Probabilistic Machine Learning @probml.bsky.social · 10/02/2026

ProbML 2026 (formerly AABI) invites submissions on probabilistic ML (both Bayesian and otherwise!), July 5 in Seoul (co-located with ICML). Website: probml.cc. Tracks: proceedings (PMLR), workshop, fast track. New focus includes applications in healthcare and climate! Submit by: 20 March 2026.

12516

Richi Paul @erpel.bsky.social · 03/02/2026

hi @auai.org, am I allowed to have wrapfigs/wraptables that span across more than one column like in the picture?

110

Richi Paul @erpel.bsky.social · 07/12/2025

I‘m presenting our work on uncertainty quantification in cell tracking later in the MedEurIPS workshop as an oral! check it out if you’re around and interested in UQ in bioimaging 🦠🔬🧠🤖

openaccess.thecvf.com

ICCV 2025 Open Access Repository

040

Richi Paul @erpel.bsky.social · 01/12/2025

congrats, cover looks amazing:)

030

Richi Paul @erpel.bsky.social · 11/06/2025

phase space looks cooler for the foward euler though :) btw, what exactly is the phase space showing? trajectories of slightly differently initialized pendulums?

100

Richi Paul @erpel.bsky.social · 11/06/2025

we found this in our institute‘s kitchen but nobody knows what it is… anybody here has an idea? fork‘s for scale

010

Richi Paul @erpel.bsky.social · 04/06/2025

nice! I was already getting concerned

010

Richi Paul @erpel.bsky.social · 14/05/2025

If a low-confidence reviewer gives a lot of criticism, which reads as if they obviously didn't get the paper, do you invest a lot of effort into correcting them and trying to clarify or do you just give up on that one and focus on the other reviews, provided you have a 1-page rebuttal limit? 😉

010

Richi Paul @erpel.bsky.social · 17/03/2025

cool, guess they‘re going to pay the reviewers now with that 500 mil yearly profit! :) lol

040

Richi Paul @erpel.bsky.social · 17/03/2025

nice, is that the latest bluesky hack to bookmark stuff?

100

Richi Paul @erpel.bsky.social · 11/03/2025

hmm, I love the rendering effect of large detailed pdf figures :)

040

Richi Paul @erpel.bsky.social · 01/03/2025

hey @iccv.bsky.social, 2 questions: am I allowed to change the font size within tables? e.g. to squeeze a rather wide table into a single column? also, are wrapfigs allowed that span across more than just 1 column like in the screenshot?

120

Richi Paul @erpel.bsky.social · 27/02/2025

at @kit.edu you have to hand in an official form, wait for an administrative notice and pay a fee just to get a document on your course of studies? pretty much sums up german bureaucracy... at @univie.ac.at I just log into my old student account and download that very same form 🫠

000

Richi Paul @erpel.bsky.social · 18/12/2024

aber 100-jahre alte bäume dafür gefällt :( und was soll überhaupt die zweie spur zwischen uni und radstreifen?

310

Richi Paul @erpel.bsky.social · 18/12/2024

having lived there for 9 yrs, I'd challenge you on that one! though it does look delicious

100

Richi Paul @erpel.bsky.social · 09/12/2024

I'll continue the rant bc it's just too dumb: so my employer provides me with a windows machine, which I barely touch, so it's still factory new. I boot it to use adobe, takes 5 minutes, everything's slow, I check the task manager -> 50% RAM due to a bunch of dell recovery and support software... 🫠

000

Richi Paul @erpel.bsky.social · 08/12/2024

ew and to top it off, the free adobe version doesn’t let me digitally sign docs? 🤢🫠😭😵‍💫

000

Richi Paul @erpel.bsky.social · 08/12/2024

how does this world run on proprietary formats?? it drives me crazy. my employer requires me to fill out pdfs w xfa forms which open source pdf viewers don’t support and adobe even though claiming otherwise don’t provide a reader for linux, but who tf works srsly with windows when u can have i3wm?

210

Richi Paul @erpel.bsky.social · 05/12/2024

saw this feed on one of my first days here on bsky and pinned it and I love it! especially once you start following a few hundred accounts, the main feed becomes too long to see all on it. bsky.app/profile/did:...

030

Richi Paul @erpel.bsky.social · 30/11/2024

hi, would love to be added as well :)

000

Richi Paul @erpel.bsky.social · 29/11/2024

me to myself at the beginning of the week, thinking about trying out a little idea I've had in my head for a while, which I hoped to have coded up within a day...

020

Richi Paul @erpel.bsky.social · 27/11/2024

haha, like you made it improper by accident and realized it has cool properties?

110

Richi Paul @erpel.bsky.social · 27/11/2024

oh, I see. well, cool paper nevertheless

110

Richi Paul @erpel.bsky.social · 27/11/2024

@lucamb.bsky.social posted a very nice paper on improper GPs the other day, guess he should be in there too :)

110

Richi Paul @erpel.bsky.social · 25/11/2024

hehe, found it: bsky.app/profile/drsc...

000

Richi Paul @erpel.bsky.social · 25/11/2024

I saw some kind of hack the other day which allowed you to basically bookmark using those customized feeds I believe? I‘d have to search for it.. Apart from that you may just like those, but it’s clearly not optimal

200

Richi Paul @erpel.bsky.social · 24/11/2024

I think in Germany the costs per year of a PhD student are estimated at about 100k€, which makes 25k PhD years, so something about 6000 PhD positions if you give them a 4-yrs contract. actually less than what I would have guessed

000

Richi Paul @erpel.bsky.social · 24/11/2024

and quite in contrary, if it wasn't about the aftertaste of trying to inflate paper counts, then I guess such only-blogpost-worthy papers are actually super nice, because in the end they're still just about making some kind of knowledge or insight a bit more persistent

100

Richi Paul @erpel.bsky.social · 24/11/2024

isn't a blog post to some extent just the modern version of a paper? personally, I don't think there's anything wrong with short papers and that any aversion against them stems mainly from the whole overvaluing of paper and citations counts

100

Richi Paul @erpel.bsky.social · 24/11/2024

I'm curious to see though how often it will be used and just generally if/how those who submitted perceive any change/improvement in review quality

120

Richi Paul @erpel.bsky.social · 24/11/2024

I guess it still makes sense then to have you on screen :)

010

Richi Paul @erpel.bsky.social · 24/11/2024

do you point on the slides? if so, I think appearing as in the screenshot is nice, otherwise maybe a bit smaller?

140

Richi Paul @erpel.bsky.social · 24/11/2024

you‘re added to my mental list of candidates for a "Junior Researchers in ML+AI" which I will turn into a starter pack once my list grows to at least 6 ppl haha, that‘s the minimum required. so if you know any other ppl here who should be added, pls link them :)

110

Richi Paul @erpel.bsky.social · 24/11/2024

we should really make a starter pack for junior ml researchers! but you need to have at least 6 such junior researchers to step sich a list, including @elnaza.bsky.social (if you want) we‘d be only 3 so far, you know any other here who might be interested?

110

Richi Paul @erpel.bsky.social · 24/11/2024

@kityates.bsky.social has two nice starter packs for sci comm :)

020

Richi Paul @erpel.bsky.social · 24/11/2024

Yeah, true, nevertheless a fair point. But again, I'd say the starter packs can and do actually help the migration, one just needs to be daring enough to ask to be included, I guess

100

Richi Paul @erpel.bsky.social · 24/11/2024

Oh, yeah, that's true. When I first got a Twitter to engage in the community I was also very intimidated. But I think being somewhat persistent about liking+commenting stuff and that getting an occasional reaction does break the ice, to the point that it becomes a nice way to randomly approach ppl

100

Richi Paul @erpel.bsky.social · 23/11/2024

I was actually just thinking about doing this... though here in Germany it's actually 1am and I just be going to bed, lol

010

Richi Paul @erpel.bsky.social · 23/11/2024

I don't necessarily agree that it's intimidating, but I do agree that the bias towards known seniors is frustrating as a junior. I post stuff, nobody replies, it sucks. But if you just ask about being part in the starter packs, you may at least get a good follower base. At least it worked for me

100

Richi Paul @erpel.bsky.social · 23/11/2024

I get the frustration, but being "a junior" myself, I do actually agree with those seniors that everything being "fresh" does help. And it's just exactly the starter packs. You just need to ask to be included there :) And so far I didn't have the impression of ppl gatekeeping those starter packs

290

Richi Paul @erpel.bsky.social · 23/11/2024

yeah, I definitely agree

100

Richi Paul @erpel.bsky.social · 23/11/2024

CVPR implemented a "irresponsible reviewer" policy this year which will result in desk rejects of the reviewer‘s submissions.. guess that’s a step? will be interesting to see how that turns out

190

Richi Paul @erpel.bsky.social · 23/11/2024

is anybody aware of any results for importance sampling, where we don't have access to the true unnormalized target density, but only to an unbiased estimator of its log density? I'm thinking e.g. about a setup where we may wanna estimate posterior densities on mini batches

000