CoLLAs is THE conference for non-i.i.d learning! If you work in lifelong/continual learning, meta-learning, in-context learning, transfer learning, multi-task learning, online learning, federated learning, unlearning, active learning, OOD generalization, open-ended learning,

I recently had the pleasure of writing a news & views article on the excellent work done by Julian Büchel et al. showing how sparse mixture of experts architecture can be leveraged on analog in-memory computer hardware. The original paper is linked in the article.

Very excited that our review of neuromorphic computing at scale has now come out. People disagree on what neuromorphic means; but one thing we all agree on about brains is that they have a lot of neurons. Getting to brain-like scales is exciting! nature.com/articles/s4158…

We are hosting & organising the UK Neural Computation Conference 2025 Imperial College London! 🚀 Fantastic set of speakers, see at neuralcomputation.uk It’s an incredibly exciting time to be working at the intersection of comp neuro/AI & applications. Save the dates: 9th July -

I'm happy to have played a tiny role in the writing of this excellent review paper talking about scaling neuromorphic computing with 22 other people. Props to Dhireesha Kudithipudi for leading the effort: nature.com/articles/s4158…

We have opened the call for application to UIUC+ Summer Research Program for undergrads: go.illinois.edu/UIUCPlus-SRP. Please apply or spread the words. The mentors are 12 UIUC PLSE professors/alumni. The (soft) deadline is February 28th, where we start reviewing the applications.

It's really disappointing to watch US media orgs consistently push out drivel about DeepSeek. The US is so obsessed with US-exceptionalism and Sinophobia that every story gets spun in the worst possible light. DeepSeek is the coolest AI start-up in the world. Write that story.

BlueSky is an alternative social media network, that is open-source and decentralised. It's super easy to set up. And no one can manipulate the content you see by tweaking the content selection algorithm 🧵

prudent reminder for scientists today, from a thousand years ago

🚀🌐Introducing AgentRxiv: a framework where autonomous research agents can upload, retrieve, and build on each other’s research. AgentRxiv takes your research direction and progressively outputs research, building on its previous work with each new paper! 🧵

Today, we're releasing ARC-AGI-2. It's an AI benchmark designed to measure general fluid intelligence, not memorized skills – a set of never-seen-before tasks that humans find easy, but current AI struggles with. It keeps the same format as ARC-AGI-1, while significantly

Unlike ARC-AGI-1, this new version is not easily brute-forced. Current top AI approaches score 0-4%. All base LLMs (GPT-4.5, Claude 3.7 Sonnet, Gemini 2, etc.) score 0%. Single-CoT reasoning models (Claude Thinking, R1, o3-mini…) score 0-1%. So you can't solve these tasks via

Economic inequality has not always been a feature of market economies. I show using new data from pre-industrial Japan, 1637-1872, that the Japanese had managed to keep inequality low for centuries without massive state intervention. How did they do it? 1/n

Lots of academics have moved to the other site, so if you’re only here you really have no idea how many serious research grants are getting outright canceled. American science will really never recover in so many ways.

My feed has been full of canceled AI grants, specifically. Safety, bias, and factual correctness have all been defunded. Digital humanities programs like Project Perseus have been cut. Training programs have ended, as have consumer studies. All the stuff that industry WON'T do.

The Third World scientist, doling out €€€, preparing a 500-page document, refreshing the VFS website for an appointment, & standing in line to donate biometric data for a Schengen visa to attend a conference, a visa that may be denied, reading « science has no passport » ...

Registration is open for UK Neural Computation 2025 Imperial College London 🚀! 10th & 11th July. Bringing together a fantastic set of speakers. We are offering up to 8 conference fellowships to support attendance at UKNC2025, sponsored by ARIA Scalable Neural Interface

katie Malcolm Gladwell is basically astrology for men. Fun to dabble with but concerning if you actually believe it

I am still beta-testing the app that helps plan science, gives feedback, digests the literature, makes todos, and write a first version of the paper. Feeling overwhelmed by the countless aspects of doing science? Try this (for free): planyourscience.com