Finnish Center for Artificial Intelligence

In the latest FCAI newsletter: ⚗ #MachineLearning for materials science, with Milica Todorović and researchers across FCAI 🤖 #GenerativeAI & productivity (Heikki Ailisto) 💡 How to cope with AI risks (Jaana Leikas) Read it all here: https://lnkd.in/d6z-jB-2 cc Helsinki Institute for Information Technology VTT Research Council of Finland | Suomen Akatemia ELLIS - European Laboratory for Learning and Intelligent Systems Aalto University Department of Computer Science University of Helsinki Laura Ruotsalainen CSC - IT Center for Science

This week at #ICML2024 [ICML] Int'l Conference on Machine Learning : We've listed FCAI's research from Aalto University and University of Helsinki in this thread, plus see below! https://lnkd.in/duZ_5T3V

Thrilled to announce a new paper in Gaussian process regression models at [ICML] Int'l Conference on Machine Learning and ELLIS Workshop on Geometry-grounded Representation Learning and Generative Modeling In this work we aim at solving the maximum a posteriori solution for a Gaussian process posterior when the log-likelihood function is strongly not concave. To do so we propose a new variant of the natural gradient and show, in an empirical example, that when the log-likelihood function strongly lacks concavity this new variant has better performance than the exact natural gradient. This is in contrast with the usual belief that the natural gradient using the Fisher metric is the go-to choice. From a Riemannian geometry viewpoint this sheds more light onto the question of how to best represent the underlying geometry of the problem Check out the full paper and see you soon in Vienna ! https://lnkd.in/dKnTYtFt

ChatGPT ja muut suuret kielimallit ovat tulleet ryminällä ihmisten tietoisuuteen. Julkisuudessa ja työpaikoilla pohditaan nyt, miten uusia tekoälymalleja voidaan hyödyntää työelämässä sekä millä tavalla ne muuttavat työtä, ammatteja ja yhteiskuntaa. Mitä jokaisen tulisi tietää tekoälystä työelämässä? Millaista työtä tekoäly voi korvata, entä mitä työtä se täydentää? Miten tekoälyn synnyttämä työelämän muutos vertautuu aiempiin teknologisiin muutoksiin? Keskustelemassa ovat historioitsija, yliopistotutkija Mona Mannevuo ja työelämäprofessori, Silo AI:n hallituksen puheenjohtajaja Tero Ojanperä sekä dekaani, tietojenkäsittelytieteen professori Sasu Tarkoma. Juontajana on Anu Partanen. Podcast on nauhoitettu Miten tekoäly muuttaa työtä? -tapahtumassa Tiedekulmassa 6.3.2024.

Registration is now open for AI Day 2024, co-located with #NordicAIMeet at the University of Helsinki on October 21-22. Check out the program and speakers: https://lnkd.in/dznAsYXR NORA – The Norwegian Artificial Intelligence Research Consortium AI Sweden WASP Research Program Pioneer Centre for AI Helsinki Institute for Information Technology Aalto University Department of Computer Science

What an event! 200 #AI researchers, #PhD students, and industry experts from the European ELLIS & ELISE community gathered in Helsinki for an in-depth conference to discuss the achievements of the #EU-funded ELISE project, explore the latest #MachineLearning research in workshops and keynote sessions, and foster new collaborations among experts in Europe! Check out this summary: https://lnkd.in/eDQY_skk ELISE is a network of artificial intelligence research hubs: https://www.elise-ai.eu Based on the highest level research, it spreads its knowledge and methods in academia, industry and society. It works in close cooperation with ELLIS, is part of the EU Horizon 2020 ICT-48 portfolio, and is coordinated by Aalto University.

Some highlights from last week's Collaborative and Generative AI Summer School at Aalto University, supported by ELISE and ELLIS - European Laboratory for Learning and Intelligent Systems 👇 👇

What's a virtual laboratory? In the collaboration between FCAI, Orion Corporation and partners, researchers are using AI methods to support early-stage drug development. Read more: https://lnkd.in/dt-th_wj cc Silo AI CSC - IT Center for Science Business Finland Admescope Pharmatest Services

AI systems are resource-intensive -- so can they ever contribute to sustainability? Our experts, including Simo Särkkä, Laura Ruotsalainen, Fabrice Saffre and guests Mats Sjöberg and Frederic Pariente, gave their opinions in a recent webinar: https://lnkd.in/dMwWd7h8 NVIDIA AI CSC - IT Center for Science LUMI supercomputer VTT

AI-driven design of materials is accelerating, with collaborations and open datasets paving the way, says FCAI professor Milica Todorović. Read more: https://lnkd.in/dJj2-4dX cc Mikko Mäkelä