Recent searches

No recent searches

Search options

Only available when logged in.
eupolicy.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
This Mastodon server is a friendly and respectful discussion space for people working in areas related to EU policy. When you request to create an account, please tell us something about you.

Administered by:

Server stats:

201
active users

eupolicy.social: AboutProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.4.5

#scicomm

21 posts19 participants0 posts today
Carlos Tomas Grahm

Video models are starting to reason — but not how you think.

arxiv.org/abs/2509.20328

Veo 3 shows that large-scale transformers can now handle video the way LLMs handle text: predicting the next “token” of motion instead of the next word.

That doesn’t mean they’ve achieved visual understanding. It means the same mathematical machinery is spreading — from words to pixels to physical processes.

There’s no new theorem here. No bold conceptual leap. But it marks something subtler: we’re finally close to unifying how machines see, speak, and simulate.

For those of us watching the frontiers, this is the stage to start paying attention — when the capabilities look underwhelming, but the framework starts to align.

(Think early deep learning papers in 2012: ordinary on paper, historic in hindsight.)

The real shift will come when these models don’t just describe motion, but discover new rules of it.

#AI #scifi #askfedi #london #Fensterfreitag #JukeboxFridayNight
#ArtificialIntelligence #MachineLearning
#DeepLearning
#VisionSystems
#GenerativeAI
#Research
#ScienceCommunication
#SciComm

arXiv logo
arXiv.orgVideo models are zero-shot learners and reasonersThe remarkable zero-shot capabilities of Large Language Models (LLMs) have propelled natural language processing from task-specific models to unified, generalist foundation models. This transformation emerged from simple primitives: large, generative models trained on web-scale data. Curiously, the same primitives apply to today's generative video models. Could video models be on a trajectory towards general-purpose vision understanding, much like LLMs developed general-purpose language understanding? We demonstrate that Veo 3 can solve a broad variety of tasks it wasn't explicitly trained for: segmenting objects, detecting edges, editing images, understanding physical properties, recognizing object affordances, simulating tool use, and more. These abilities to perceive, model, and manipulate the visual world enable early forms of visual reasoning like maze and symmetry solving. Veo's emergent zero-shot capabilities indicate that video models are on a path to becoming unified, generalist vision foundation models.
GrrlScientist ⧖ Ⓥ 🇺🇦

A River’s Microbiome May Protect Wild Salmon Against Malnutrition

"Chinook salmon may be canaries in the coal mine by revealing that widespread thiamine deficiency could be quietly chiseling away at fish, bird, and wildlife populations across the northern hemisphere"

#SciComm by @grrlscientist

#microbiome #rivers #salmon #ChinookSalmon #thiamine #vitaminDeficiency #malnutrition #ecology #conservation medium.com/grrlscientist/a-riv

MPI for Gravitational Physics

Join Karsten Danzmann, director at the @mpi_grav in Hannoer and professor at @unihannover, for a talk about the space-based gravitational-wave detector LISA.

As part of the space exhibition “Mars, Moon, and Gravity” from 16 to 30 October 2025, at the A2 Center near Hannover, there will be an entertaining evening lecture event on 24 October at 18:30.

Karsten Danzmann will talk about LISA, the planned space observatory for gravitational waves which will revolutionize our understanding of the dark, invisible side of the Universe starting in the 2030s.

Franz Renz, professor at @unihannover will discuss robotic missions to Mars, our nearest neighbor in space, and how research in Hannover contributes to them.

#Hannover#Exhibition#Space
Sci_Burst

Come with us to #floriade as science communicators (part 1)👩‍🔬💐

This year’s theme was “Science and Nature” ! Each garden bed display showed a different #science element, from #Petri dishes 🧫 to #DNA 🧬 to the human #brain 🧠 to light #refraction 🌈 to #space and #engineering mechanisms ⚙️🪐 and so much more. Not to mention all the interactive science-themed shows and activities 💜

Flowers themselves are all about #biology and #genetics 🧪🌷

#scicomm#floriade2025#floriadeaustralia
Continued thread
*
Alexander Knochel

Yang-Mills theory famously has the property that, unlike the photon in electrodynamics, it is forbidden for its bosons to possess any mass - it would be much easier to bestow mass upon the photon of electrodynamics via the Stückelberg mechanism (named after none other than Ernst Carl Gerlach Stückelberg von Breidenbach zu Breidenstein und Melsbach, unsung hero of quantum electrodynamics), but such shortcuts of generating mass are not possible in Yang-Mills-theory without seriously breaking things.

According to legend, this led to a historic clash between the Yang-Mills-team and acerbic father of quantum mechanics and pitiless axe of reason, Wolfgang Pauli, who let Yang-Mills-theory die from public humiliation after a talk by remarking that it predicts massless bosons which are clearly not present in nature.

What Wolfgang Pauli could not have known - the simplest and most elegant way to have masses for bosons in Yang-Mills-Theory, it turned out later, is the Higgs-mechanism, and the rest, as you know, is science history...

#science#physics#scicomm
*
Alexander Knochel

Rest in Peace, Frank Yang!

I had the privilege to get my masters at his institute at SUNY that bears his name, although I did not meet him during my time there as he was visiting China at the time - my claim to Yang fame thus remains that I got to fill out my tax paperwork sitting in his office, and I think we stole one issue of CERN courier addressed to his name ;)

While he got his Nobel prize for the prediction of parity violation, the context in which quantum field theorists and particle physicists will probably encounter his name most often is for Yang-Mills-Theory, the mathematical framework which describes the strong and weak interactions in the standard model of particle physics, named after Yang and Robert Mills who unfortunately died 26 years ago..

If you don't know it, you can imagine Yang-Mills-Theory as having several copies of the electromagnetic field ( eight in the case of the strong interactions, and three in the case of weak interactions) in parallel, and have them all interact with each other. There is a unique way to achieve this in four spacetime dimensions which allows for the dynamics observed in Gluons and the W, Z bosons with the photon, and that is Yang-Mills-Theory.

It would be fair to say that Yang-Mills-Theory is the foundation of how we understand most of the fundamental forces in modern physics.

#science#physics#scicomm
TrendingLive feeds
About