LAB42 UvA

🪐 Join us for an exciting Hackathon on Equivariant Neural Fields (ENFs) & Geometry-Grounded Representations! ✨ The Deep Thinking Hour, ELLIS unit Amsterdam, and the New Theory joint forces, and hosted a hackathon focused on Equivariant Neural Fields (ENFs) and geometry-grounded representations 🌆 Programs of the day 🔹 Dr. Erik Bekkers (University of Amsterdam) will dive into geometry-grounded representations and his vision for the future of representation learning 🔹 Christian Shewmake, CEO of New Theory, will explore advances in foundation models—and what’s still missing 🔹 David Knigge and David Wessels (University of Amsterdam) will introduce their cutting-edge research on ENFs and breakdown the hackathon details 🍲 Then, we fuel up with a free lunch before diving into the challenge! 💡 The Hackathon Challenge: The hackathon focuses on developing new inference or encoding schemes based on ENFs. ENFs try to overcome texture bias in CNNs and ViTs by grounding representations in geometry. Currently, geometry-grounded methods still struggle to create meaningful positional features and, at the same time, be as expressive as Sota models such as CNNs and standard Transformers. 🎯 Your mission: With this hackathon, we want to introduce you to ENFs and challenge you to try figure out solutions for new encoding schemes resulting in meaningful positional features. We will work in teams on the hackathon and have a benchmark to evaluate. 📆 Thursday, 27 February 2025 ⏰ 10:00-17:00 📍 L3.36 at LAB42 UvA 📢 Sign up now, the spots are limited! https://lnkd.in/ewqNGRkw #ELLISunitAmsterdam #deeplearning #geometricdeeplearning #AI #ML

President of the Executive Board of the UvA: "⬇️ The focus is on #AI and #Quantum computing."

LAB42 resident Zeta Alpha! ⬇️ "Other promising AI companies [...], as well as Zeta Alpha, founded by Jakub Zavrel, which develops AI models that enable other companies to analyse their own confidential business information. https://lnkd.in/d4pf5HBW

Trying something new and exciting: I’ll be teaching a Masterclass for Professionals on Game Theory and Market Design at UvA this June. Registration now open. Great if you can repost! 🙏 🔗 https://lnkd.in/eyDe5u2d #lifelonglearning #marketdesign #mechanismdesign #gametheory #speltheorie #strategicthinking #strategischdenken #auctions #veilingen #matching #centraleloting #algorithms #AI #amsterdam LAB42 UvA - UvA Faculty of Science - UvA Academy

Upcoming events @ LAB42 ⬇️ More information: lab42.uva.nl/events.

Last week, professionals from the UvA Teaching & Learning Centre visited LAB42 as part of their event program. They explored questions such as "How do we make curricula future-proof?" and "How do we prepare students for AI?" During their visit, they attended a demo by the Dutch Nao Team of robots playing football at the Intelligent Robotics Lab. After that the group engaged in discussions with members of research groups and companies, like Professor of Data Science Paul Groth Groth and Daan Geijs from kaiko.ai. Thanks for visiting us, we were glad to have you!

Thanks for visiting us as well!

SignLab Amsterdam, part of the University of Amsterdam, is working with Koninklijke Auris Groep on an app to support parents of deaf and hearing-impaired children in learning Dutch Sign Language (NGT). A ZonMw grant (innovation in health, care and well-being) was recently awarded for the development of this app. The app ‘ZininNGT’ helps parents learn more sentences in NGT faster. This strengthens the (language) development of deaf and hearing-impaired children. The collaboration between the UvA and Auris contributes to a more inclusive society. The innovative approach and technologies of this project also potentially reach further than the Netherlands. They can serve as an inspiration and model for similar initiatives worldwide, which will further increase the impact. This innovative project is led by Floris Roelofsen, professor of Formal Semantics at the Institute for Logic, Language and Computation (ILLC) - University of Amsterdam (UvA) and director of SignLab Amsterdam. Part of the project is the PhD research of Jos R. Ritmeester, who is investigating the development and use of the app. The Auris innovation team is involved in the development of the app and the Research team in the supervision of the PhD research.

📢 The Computational Linguistics Seminar series: the Interplay between Language and Reasoning is scheduled for Thursday, February 6th, 2025 (16:30) and will feature Raffaella Bernardi, University of Trento ✨ 📍 It will take place in Room L0.06 at LAB42 UvA, AmsterdamScience Park, with live streaming on Zoom 🌍 Abstract Large Language Models, and ChatGPT in particular, have recently grabbed the attention of the community and the media. Having reached high language proficiency, attention has been shifting toward its reasoning capabilities. It has been shown that ChatGPT can carry out some simple deductive reasoning steps when provided with a series of facts out of which it is tasked to draw some inferences. In this talk, I will argue for the need for models whose language generation is driven by an implicit reasoning process and a communication goal. To support my claim, I will present two papers recently produced within my group: one evaluates LLMs’ formal reasoning skills (Bertolazzi et al., 2024) and the other focuses on LLMs’ information-seeking strategies (Mazzaccara et al., 2024); to this end, we take syllogisms and the 20-Questions game as test beds. These tasks have been used extensively in cognitive sciences to study human reasoning skills, hence they provide us with a variety of experiments to inspect the language and reasoning interplay in LLMs. 🔗 For more information, please visit: https://lnkd.in/e4Htf7ss

Meet professor Raquel Fernández: ‘My research group is working on capturing the level of uncertainty of a computational model and figuring out the best way for the model to express that. It is very important to give that information to the user so that the technology can be trusted.’