✨️ Attention!

How would one protect private datasets and train on them? Only federated learning and differential privacy might not be enough 🤔

#ArtificialIntelligence affects us all! At ELLIS, we are creating a multicentric research lab attracting talent from around the world to secure Europe’s leadership in this crucial field. Take a look at our new video and #JoinELLISforEurope! youtube.com/watch?v=euqOfQ…

Should be very interesting!

Proud of my collaboration with EPFL, Maksym Andriushchenko and Martin Jaggi in particular 😎

If you want to visit me at Know Center to work in the fields of information theory for deep learning or physics- informed neural networks, this might be a great opportunity. Drop me a line if you are interested!

What is the correct granularity of the world around us? 🤔

We are presenting three works at ICLR next week: 1️⃣ Layer-wise Linear Mode Connectivity (main track): some curious explorations of the layer-wise structure of deep net loss surfaces. Includes experiments on ConvNets, transformers, Pythia LLMs, connections to federated learning

One more perspective to tackle understanding of loss surfaces 🤓: adversarial examples!

Stop discarding your old gradients! Introducing AdEMAMix, a novel (first-order) optimizer capable of outperforming Adam. Let’s have a thread on momentum and the surprising relevance of very old gradients. A joint work with David Grangier and Pierre Ablin #ml #optimization 1/🧵

I was happy to present the initial results from the research about information-theoretic view on generalization, conducted together with Bernhard Geiger in the frame of the #enfield grant. The #enfield second call is already open! enfield-project.eu

The Road to AGI along with Emiliano (who's awesome, go follow), I built an interactive timeline of everything in AI the past few years we're living through the most exciting time in history and this site hopes to document it! go visit: ai-timeline dot org (link below)

I wonder if it is common place, that after finishing doctorate in AI one really cannot find a job: too little practical experience for industry, not fancy enough for big research companies. Being rejected just based on your CV is somehow frustrating.

Being part of #ELLIS was one of the best experiences in my PhD path. I hope to continue productive collaboration with the network further as a more senior researcher!

🚨 Incredibly excited to share that I'm starting my research group focusing on AI safety and alignment at the ELLIS Institute Tübingen and Max Planck Institute for Intelligent Systems in September 2025! 🚨 Hiring. I'm looking for multiple PhD students: both those able to start

Keep attending! Now our meetings will get even more interesting discussions