Serge Belongie
banner
serge.belongie.com
Serge Belongie
@serge.belongie.com

Professor, University Of Copenhagen 🇩🇰 PI @belongielab.org 🕵️‍♂️ Director @aicentre.dk 🤖 Board member @ellis.eu 🇪🇺 Formerly: Cornell, Google, UCSD

#ComputerVision #MachineLearning

Serge Belongie is a professor of Computer Science at the University of Copenhagen, where he also serves as the head of the Danish Pioneer Centre for Artificial Intelligence. Previously, he was the Andrew H. and Ann R. Tisch Professor of Computer Science at Cornell Tech, where he also served as Associate Dean. He has also been a member of the Visiting Faculty program at Google. He is known for his contributions to the fields of computer vision and machine learning, specifically object recognition and image segmentation, with his scientific research in these areas cited over 150,000 times according to Google Scholar. Along with Jitendra Malik, Belongie proposed the concept of shape context, a widely used feature descriptor in object recognition. He has co-founded several startups in the areas of computer vision and object recognition. .. more

Computer science 91%
Engineering 7%

Reposted by Serge Belongie

Next we would like to acknowledge our two Platinum Sponsors @novonordisk.bsky.social and @sprind-de.bsky.social (who will be making a big announcement at #EurIPS!) for their major contribution to the #EurIPS conference.

Make sure to visit our sponsors during the conference to learn more about them!

"He had joined @caltech.edu's #Visipedia project aimed at training computers to recognize objects in photos, using machine learning." (2/2)

Merlin's journey from 20 Questions to Photo ID to Sound ID, featuring Grant Van Horn, "an undergrad at @ucsandiego.bsky.social [circa 2010] knocking on doors and looking for professors to work with." (1/2)
Ever felt the magic of Merlin? From spotting new birds to learning their songs, Merlin connects millions worldwide. What started as an idea in 2008 at Cornell Lab is now a global movement uniting people with birds, one sighting at a time. Full story: www.birds.cornell.edu/home/the-mag...

Reposted by Serge Belongie

OVER 500 #EurIPS WORKSHOP POSTERS 🎉🎉🎉

We are exceedingly happy to share that our 18 EurIPS workshops have seen an immense amount of interest, meaning that over 500 posters will be presented on the 6th and 7th December.

More workshops details at eurips.cc/workshops/
Workshops - A NeurIPS-endorsed conference in Europe
A NeurIPS-endorsed conference in Europe held in Copenhagen, Denmark
eurips.cc

Reposted by Serge Belongie

Excited to share our paper Representational Difference Explanations (RDX) was accepted to #NeurIPS2025! 🎉RDX is a new method for model diffing designed to isolate 🔍 representational differences. 1/7

Fascinating and insightful talk by @giadapistilli.com on Human-AI Companionship, presented at today’s @ellisinstitute.fi launch event in Helsinki

One of the key findings is that AI chatbots fail to set boundaries when users are most vulnerable, e.g., expressing love or loneliness

Reposted by Serge Belongie

🚀 ELLIS Institute Finland 🇫🇮 officially opens today, launching a hub for cutting-edge AI research, talent development, and collaboration across academia, government & industry.

Opening remarks by @samikaski.bsky.social, @serge.belongie.com, Bernhard Schölkopf, PM Petteri Orpo & Peter Sarlin

Here’s the latest example. It’s the people, stupid. If you must anthropomorphize the AI, then consider it the useful idiot. It does not plot. It does not deceive.
Exclusive | Chinese Hackers Used Anthropic’s AI to Automate Cyberattacks
The use of AI automation in hacks is a growing trend that gives hackers additional scale and speed
www.wsj.com

How it started / How it’s going

See you in Copenhagen! 🇩🇰

eurips.cc

#EurIPS #NeurIPS

Reposted by Serge Belongie

Come work with my team in Sweden on machine learning for remote sensing and urban climate resilience! www.ri.se/sv/om-rise/j...

Cool new work from UoE, Cornell, and UMass evaluating models of human visual learning
From medicine to geo-guessing, humans can get incredibly good at solving visual recognition tasks.
But how is this skill learned, and can we model its progression?
We present CleverBirds, accepted #NeurIPS2025, a large-scale benchmark for visual knowledge tracing.
📄 arxiv.org/abs/2511.08512
1/5
CleverBirds: A Multiple-Choice Benchmark for Fine-grained Human Knowledge Tracing
Mastering fine-grained visual recognition, essential in many expert domains, can require that specialists undergo years of dedicated training. Modeling the progression of such expertize in humans rema...
arxiv.org

Reposted by Serge Belongie

From medicine to geo-guessing, humans can get incredibly good at solving visual recognition tasks.
But how is this skill learned, and can we model its progression?
We present CleverBirds, accepted #NeurIPS2025, a large-scale benchmark for visual knowledge tracing.
📄 arxiv.org/abs/2511.08512
1/5
CleverBirds: A Multiple-Choice Benchmark for Fine-grained Human Knowledge Tracing
Mastering fine-grained visual recognition, essential in many expert domains, can require that specialists undergo years of dedicated training. Modeling the progression of such expertize in humans rema...
arxiv.org

The problem is us, with our Paleolithic vulnerabilities, our FOMO, our susceptibility to snake oil salesmen and the ELIZA effect. Say no to anthropomorphized tech solutionism and yes to stronger human institutions, fortified by ordinary technology. (11/11)

AI-based tools — even terribly unsophisticated ones — have dramatically lowered the barrier of entry for bad actors to do harm. (10/11)

The fundamental problems remain. Bad software leads to bad outcomes. Phishing, grifting, misinformation, scams, and other con jobs threaten us from all sides, like nitrogen runoff encroaching upon a lake. (9/11)

If we, as a society, hand the keys to agentic software, and erode the classic lines of defense that humans have built up for decades in cybersecurity and institutional best practices, then we deserve whatever sad apocalypse ensues. (8/11)

As Anders Søgaard noted, Animal Behavior researchers have wrestled with the anthropomorphism challenge, and so should we. (7/11)

(2) “AI minus Bullshit equals Ordinary Technology.” We need to stop using anthropomorphic terms when we talk about AI. Yes, AI researchers themselves started it, going back decades, but times change, and we need to correct course. (6/11)

With that foundation in place, the use of AI as a tool in capable human hands could lead to abundance in various sectors, thereby building up defenses against some of our less desirable Stone Age impulses. (5/11)

On the bright side, I think there is a lot we can do in the second category: we can build better institutions. We can build them upon a foundation of trust, safety, and functioning democracy. (4/11)

What can we do about this? With regard to the first one, I would say: not much. For the last one, as scientists, we can do our best to demystify technology — while preserving a sense of awe and wonder — but the forces against this appear insurmountable. (3/11)

(1) E. O. Wilson once remarked that “the real problem of humanity is the following: we have Paleolithic emotions, medieval institutions, and godlike technology.” (2/11)

Here are a couple of the points I made during yesterday’s Algoritmer, Data og Demokrati - ADD projektet Partnerkonference (Link: algoritmer.org/aktiviteter/...) (1/11)

Reposted by Serge Belongie

AND WE ARE FULLY SOLD OUT!

We have now sold out, with 3 weeks still to go before #EurIPS kicks off. We are looking immensely forward to seeing you all in lovely Copenhagen, sharing and exploring top-tier AI research in Europe.

Link to application portal:
Harnessing the power of AI for biodiversity monitoring with camera trap networks - From foundation model to edge processing
phd.tech.au.dk

Prof. @tokehoye.bsky.social (Aarhus University) and I have an open PhD position (jointly advised) on biodiversity monitoring with camera trap networks. Deadline: 15-Jan-2026

Please help us share this post among students you know with an interest in Machine Learning and Biodiversity! 🤖🪲🌱

Reposted by Serge Belongie

The #EurIPS Salon des Refusés posters are now online!

If you want your rejected NeurIPS paper featured, dont worry! You can still apply, but posters will be accepted on a rolling basis until the session is filled or Friday the 14th Nov, whichever comes first.

eurips.cc/salon-des-re...
Salon des Refusés - A NeurIPS-endorsed conference in Europe
A NeurIPS-endorsed conference in Europe held in Copenhagen, Denmark
eurips.cc

Doomers, Bloomers, Deemers, Bleemers

Read about Søgaard’s AGI matrix in our new CAISA Research Brief: Artificial General Intelligence and its Societal Implications caisa.dk/forskning

Interview from last week’s AI in Science summit in Copenhagen ais25.eu
Europe’s Win and Worry with Artificial Intelligence
Launch of RAISE – the Resource for AI Science in Europe.
www.sciencestories.dk