EPFL Applied Machine Learning Days: Energy, Maturity and Optimism

Published

6 May, 2024

As we look ahead to the developments in AI that will feature at EPFL’s Applied Machine Learning Days 2025 (AMLD 2025), we reflect on this year’s event with a focus on safe and transparent AI for society.

It may sound like a cliché but, as almost 2000 delegates from 40 countries participated in sessions from AI in drug discovery to how machine learning can accelerate climate action, it was clear at this year’s AMLD that the AI revolution has taken off.

Little more than a year after ChatGPT was unleashed on the world, the explosive growth in the use of generative AI tools and foundational machine learning models shows no let up and this game changing technology is already transforming the ways we live and learn.

Showcasing machine learning applications from healthcare and sustainability to neuroscience and democracy, EPFL’s AMLD is one of the largest machine learning events in Europe, bringing together experts from industry, academia, and the public sector with a focus on the practical application of artificial intelligence.

“Machine learning is a booming technology and AI is now everywhere. AMLD is truly at the intersection of cutting-edge science and the application of this technology and that’s hard to find elsewhere. I think AMLD’s uniqueness is in bringing together the science and the research with industry and policy makers,” said Prof. Marcel Salathé, co-founder of the event and co-director of the EPFL AI Center.

To kick off the event, 28 different workshops gave participants the opportunity to hear the latest insights into specific sectors and topics within machine learning, including the certification of trustworthy AI systems, probabilistic forecasting for planning, how machine learning is being used to support the UN’s Global Goals and AI in architecture. Other workshops offered hands-on training on the applied ML skills needed in different sectors.

More than 40 tracks showed how the latest deep tech and foundation models are impacting industry, science and engineering in finance, machinery, physics , oncology, and many other domains. Keynote speakers included Timothee Lacroix, co-founder ofMistral AI, who showed that the essence of impactful AI lies in the quality of data and Aleksander Madry, Head of Preparedness at OpenAI who reminded the audience that it’s up to us to be prepared to ensure that AI is used for the greater good.

EPFL School of Computer and Communication Sciences alumni, Michele Catasta the VP of AI start-up, Replit gave another keynote discussing the company’s latest advances in AI and the importance of AI in software development, as Replit is trying to create the 1000x software developer using AI.

Whilst Catasta is based in Silicon Valley he explained why he thought it was important to be at EPFL’s AMLD, “I’m European but it feels to me that most AI discussions revolve around Silicon Valley and I believe it’s extremely important for AI to be decentralized. Switzerland is in a very strategic position in Europe to help develop as much AI talent as possible globally and the list of esteemed guests that AMLD has had through the years is nothing short of amazing. I hope this event keeps growing and thriving in the years to come.”

Showcasing a health-care breakthrough, Assistant Professor Mary Anne Hartley, head of the Laboratory for intelligent Global Health Technologies, hosted jointly in EPFL’s Machine Learning and Optimization Laboratory and the Yale School of Medicine, presented MEDITRON, a 70-billion parameter open-source Large Language Model designed to help guide clinical decision-making.

MEDITRON’s release aligns with the mission of the EPFL AI Center which puts an emphasis on how responsible and effective AI can advance technological innovation for the benefit of all sectors of society and it’s this focus that Hartley believes makes AMLD so unique.

“I was in Nairobi a few weeks before for AMLD Africa, and it was so special to see how EPFL can have a reach in low resource settings, to rally people together around these kinds of topics and to spread knowledge in places that typically don’t have access to it. It really mimicked something of the magic that happens in Lausanne. AMLD is quite simply a showcase for so much interesting work – it’s one of the few places where you can take a real sneak peak at the future.”

AMLD is spearheaded by the EPFL AI Center and the EPFL School of Computer and Communication Sciences (IC). Join the next edition in Lausanne: February 11 to 14, 2025.

Author: Tanya Petersen

Source: EPFL