Lightnews — Scholar-powered news

Gergely Neu @neu-rips.bsky.social · Sep 8

check out this PhD opportunity for female students from Africa at @upf.edu :
www.upf.edu/web/phd-engi...
i have hired my amazing student @NnekaOkolo4 via this program, which was a great success. hope i can find someone this time as well!
let me know if you have any questions

Dorcas Muthoni PhD Fellowship - PhD in Information and Communication Technologies - UPF

www.upf.edu

3 6

Reposted by Gergely Neu

Timnit Gebru @timnitgebru.bsky.social · Aug 22

Lol, we need the "AI" that is killing the environment and guzzling untold amounts of resources to save the environment. This is 1984 type of marketing, like "ministry of peace" that is actually the war department.

Nature @nature.com · Aug 22

Without artificial-intelligence technologies, balancing human-caused greenhouse-gas emissions with carbon removals by 2050 is out of reach

go.nature.com/41LtWWk

Net zero needs AI — five actions to realize its promise

Without artificial-intelligence technologies, balancing human-caused greenhouse-gas emissions with carbon removals by 2050 is out of reach. Action in five areas is needed to keep this goal alive.

go.nature.com

17 240 690

Gergely Neu @neu-rips.bsky.social · Aug 8

just kidding --- i'm happy to stay ignorant until i get back from vacations in september!

4

Gergely Neu @neu-rips.bsky.social · Aug 8

i wish i was there so that i could know :p

1 2

Reposted by Gergely Neu

Ben Recht @beenwrekt.bsky.social · Aug 4

I’m excited to announce that my new book, _The Irrational Decision_, is now available for pre-order from Princeton University Press.

The Irrational Decision

How the computer revolution shaped our conception of rationality—and why human problems require solutions rooted in human intuition, morality, and judgment

press.princeton.edu

9 30 120

Reposted by Gergely Neu

da share z0ne @dasharez0ne.bsky.social · Jul 28

NEVER !! - dashare.zone ADMIN

A SKELOTON WHO FEEL'S LIKE MACREADY IN DA 1982 FILM "THE THING" AND HE WAS RIGHT TO DESTROY CHESS WIZERD!!! HIS ONLY MISTAKE WAS WASTING WHISKEY !!!!! NONE OF US COULD HAVE SEEN "THE THING" COMING BECUASE WHO WOULD KNOW THAT YOUR HANDS WOULD GO INSIDE AND IT WOULD MUNCH THEM OFF BUT WE ALL KNOW THE FEELING OF DESTROYING A COMPUTER , WE ALWAYS KNEW THEY WERE FUCKED , WERE LUCKY WE GOT A COUPLE GOOD DECADE'S OUT OF THEM BEFORE IT ALL TURNED TO SHIT!!! IMAGINE LOOKING AT YOUR FELLOW HUMAN'S AND THINKING OF EACH ONE IN YOUR MIND AS A HURDLE THAT STAND'S BETWEEN YOU AND MORE MONEY , IMAGINE A LACK OF AESTHETIC SO DEEP THAT YOU CANNOT TELL A COMPUTERS OUTPUT FROM A HUMAN BEING, IMAGINE A MIND SO DEVOID OF EMOTION THAT YOU CAN SAY WITH A STRAIGHT FACE THAT ART MUST NO LONGER BE MADE, THAT CHAINS AROUND THE WRISTS OF HUMANITY IS THE WAY IT MUST BE, YOUR OWN NASTY MIND TALKINT TO ITSELF LIKE HANNIBAL LECTER AS YOU DREAM OF A WORLD WHERE YOU CAN GET OUT OF PAYING JODIE FOSTER, WHAT EVER HAPPENED TO IMPRESSING HER LIKE WE USE TO , ANY WAY THE TEXT " I WILL NOT LISTEN TO AI MUSIC , I WILL NEVER WATCH AI VIDEO, I WILL NEVER READ AI WRITING , I HATE CMOPUTERS NOW, THEY RUINED GOING ON THE COMPUTER" - DASHARE.ZONE ADMIN

620 1.9K

Gergely Neu @neu-rips.bsky.social · Jul 28

wow 😳

Gergely Neu @neu-rips.bsky.social · Jul 28

1 9

Gergely Neu @neu-rips.bsky.social · Jul 28

well then i'm just gonna write "imagine two curves with one going down to zero way faster than the other one --- yes the faster one is ours", that cannot be illegal

1 6

Gergely Neu @neu-rips.bsky.social · Jul 28

dear reviewers, of course that "no pdf" rebuttal policy is not going to stop me from running all those wonderful new baselines you just suggested, please find the results below

3 36

Reposted by Gergely Neu

Ulrike Luxburg @ulrikeluxburg.bsky.social · Jul 21

I love the #eurips initiative! But to live up to its potential, it should be accepted an official #neurips conference location (similar to Mexico City) and not just an addon! Then we would save C02, rather than adding to it!!! What can we do to achieve this?

EurIPS Conference @euripsconf.bsky.social · Jul 16

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅

EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu

eurips.cc

1 20 74

Gergely Neu @neu-rips.bsky.social · Jul 16

what's the opposite of FOMO called again? either way, i'm right in the middle of that
(hope everyone else is doing fine at #ICML2025 nevertheless!)

15

Gergely Neu @neu-rips.bsky.social · Jul 16

great initiative -- I'll be there!

EurIPS Conference @euripsconf.bsky.social · Jul 16

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅

EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu

eurips.cc

6

Reposted by Gergely Neu

NeurIPS Conference @neuripsconf.bsky.social · Jul 16

NeurIPS is endorsing EurIPS, an independently-organized meeting which will offer researchers an opportunity to additionally present NeurIPS work in Europe concurrently with NeurIPS.

Read more in our blog post and on the EurIPS website:
blog.neurips.cc/2025/07/16/n...
eurips.cc

eurips.cc

A NeurIPS-endorsed conference in Europe held in Copenhagen, Denmark

eurips.cc

2 38 120

Reposted by Gergely Neu

da share z0ne @dasharez0ne.bsky.social · Jun 29

THINK ABOTU IT - dashare.zone ADMIN

A SKELITON AT WORK AND THERE THINKEN ABOUT HOW A.I IS FORCED INTO EMAILS, INTO SEARCHES, INTO ALL APP'S, AND IT NEVER MAKES ANYTHING BETTER, IT MAKES THERE JOB WORSE, AND DA TEXT SAYS "IF AI WAS ACTUALLY USEFUL THEN THEY WOULDNT HAVE TO FORCE US TO USE IT" AND THERE RIGHT TO SAY IT, SOMEONES MAKEN SOME MONEY OFF THIS AND IT SURE AS HELL AINT US - DASHARE.ZONE ADMIN

850 2.6K

Gergely Neu @neu-rips.bsky.social · Jun 8

nooo

1

Gergely Neu @neu-rips.bsky.social · May 30

and by "discuss" i mean discuss with each other because i'm obviously off to sleep now, good night!

5

Gergely Neu @neu-rips.bsky.social · May 30

for the first time since forever, today i felt like a "normal" person should feel like on a friday afternoon. chilled out, not worried about deadlines, todo lists, inbox etc.

all it took was staying up until 5am last night to finish a grant proposal 😂😂

#ADHD or #academia? discuss!

2 10

Gergely Neu @neu-rips.bsky.social · May 27

wanna know how to do inverse Q-learning right? read this paper then!!
joint work with the best team of students ever ♥️

Antoine Moulin @antoine-mln.bsky.social · May 27

new preprint with the amazing @lviano.bsky.social and @neu-rips.bsky.social on offline imitation learning! learned a lot :)

when the expert is hard to represent but the environment is simple, estimating a Q-value rather than the expert directly may be beneficial. lots of open questions left though!

2 15

Gergely Neu @neu-rips.bsky.social · May 26

thank you! we'll see about the prominence later, for now it's mostly about the fun :)

Gergely Neu @neu-rips.bsky.social · May 26

i generally think this work uncovers a space of ideas with huge potential impact, and many beautiful intellectual challenges to address.
check out the paper for more details if you also find it interesting:
arxiv.org/abs/2505.18005
12/12

Distances for Markov chains from sample streams

Bisimulation metrics are powerful tools for measuring similarities between stochastic processes, and specifically Markov chains. Recent advances have uncovered that bisimulation metrics are, in fact, ...

arxiv.org

4

Gergely Neu @neu-rips.bsky.social · May 26

i learned a lot while working on this project with this amazing team, and am very happy with the results! we have tons of open questions and ideas for future research, mostly about introducing function approximation and making the algorithm scalable.
11/n

1 1

Gergely Neu @neu-rips.bsky.social · May 26

we also have some cool model-selection experiments that demonstrate the ability of our approach for computing similarities between Markov chains, which could be potentially very useful for representation learning in RL
10/n

2 1

Gergely Neu @neu-rips.bsky.social · May 26

interestingly, the variables λ_X and λ_Y can be interpreted as "encoder-decoder" pairs giving the conditional distributions of Y|X and X|Y.
we ran some experiments to illustrate this, e.g., here we use SOMCOT to discover the latent structure of a block Markov chain without any prior knowledge
9/n

1 3

Gergely Neu @neu-rips.bsky.social · May 26

the algorithm is easy to implement and comes with nice sample-complexity guarantees. notably, no "concentrability" / "coverage" assumptions are required for the state distributions, the guarantees only depend on the sizes of the two chains
8/n

1 2