Now accepted @icmlconf as Spotlight!
The connection between DRO and imprecise probabilities has long felt like folk knowledge: Everyone sort of knew they rubbed shoulders, but making that intuition precise was a real pleasure.
Catch with Mengqi in Seoul if you're interested!
How do we make worst-case decision-making meaningful when test-time data can contain arbitrary perturbations — without assuming the space and loss are bounded?
We propose 𝗯𝘂𝗹𝗸-𝗰𝗮𝗹𝗶𝗯𝗿𝗮𝘁𝗲𝗱 𝗰𝗿𝗲𝗱𝗮𝗹 𝗮𝗺𝗯𝗶𝗴𝘂𝗶𝘁𝘆 𝘀𝗲𝘁𝘀 https://t.co/4y9OvbaMWI 1/5
The SIPTA seminars are back, and we are delighted to have Marnix Suilen discuss partially observable Markov decision processes and their connections with imprecise probabilities! Join us on May 6 at 15:00 CET. All the information at https://t.co/HUHhRTuCxc
New paper in The Inaccessible Game series:
Starting from a small tautology — a self-adjudicating system can’t seek external arbitration — I arrive at a “No Barber Principle” for avoiding Russell-style paradoxes.
It connects entropy, copying, and Lawvere
https://t.co/0duIzGhkQl
I’ll be in Lille next Thursday, April 23, presenting Imprecise Probabilistic ML, and our work on Conformalized Credal Regions for Classification w/ Ambiguous Labels (https://t.co/CZK8LXyPpi) at Batiste Le Bars’ MAGNET Lab, @Inria_Lille. If you’re around, come say hi!
🏹 Job alert: Dame Kathleen Ollerenshaw Fellow AI & Scientific Computing in Regulatory Science
📍 Manchester
⏰ Apply by April 15th
🔗 https://t.co/ZVfeIt0LZf
🏹 Job alert: Research Associate Multimodal AI for Translational Medicine at University of Manchester
📍 Manchester 🇬🇧
⏰ Apply by 15 April
🔗 https://t.co/YQXZAP1dil
New. The Rhetoric of Machine Learning. It really does feel that lots of us are moving in a similar direction. Thought provoking.
H/T to @mic_caprio
https://t.co/LFYiqzhqh4
Join our 2nd Workshop on Epistemic Intelligence in Machine Learning at @icmlconf to explore theoretical foundations, algorithmic innovations, or real-world applications that grapple with "unknown unknowns"
📢 Call for papers: https://t.co/aIkluoLTDe
Next Monday I’ll be presenting our work on Robust Random Features Models (https://t.co/IiqmViSc1Q) at @LauchLab at @ucl. Looking forward to the discussion; if you’re in London, let’s have a chat!
What is epistemic intelligence and why does it matter for AI safety, alignment, and continual learning?
We’re recruiting reviewers for EIML @ ICML2026 (Seoul).
👉 https://t.co/kuXqQTx4J4
Join us & please share! @icmlconf
We're pleased to be one of the co-sponsors of the 27th Postgraduate Group Theory Conference, an annual event for PhD students in group theory and related areas.
This year, it's taking place at @OfficialUoM from 30 June - 3 July.
Further details ➡️ https://t.co/OSJBfT77R3
Math is called a universal language, but history shows its global celebrations are rarely "borderless." This article into past U.S. hosted ICMs provides context for the upcoming Philadelphia 2026 ICM. I’m devastated that world politics is once again intertwined with the celebration of world mathematics. Sharing this so everyone can see the historical patterns and make up their own minds. 🌐📐
https://t.co/PVnOninVDj
#ICM2026 #Mathematics #STEM #Philadelphia