UNIVERSITA DELLA SVIZZERA ITALIANA

Their core work

USI is a Swiss university in Lugano with strong research groups in computational science, machine learning, and software engineering. Their core strength lies in advanced mathematical and computational methods — from molecular dynamics simulations and metadynamics to geometric deep learning on graphs and manifolds. They also maintain active research in software quality tools, display perception optimization, and contribute to European biomonitoring and epidemiological studies. Their work bridges fundamental algorithmic research with practical applications in areas like fake news detection, drug binding prediction, and high-performance computing.

What they specialise in

How they've shifted over time

In their early H2020 period (2015–2018), USI focused on foundational machine learning — geometric deep learning, social network analysis — alongside public engagement (SPARKS) and molecular simulation (VARMET). From 2019 onward, the portfolio shifted toward applied computational science: high-performance computing, exascale methods, software engineering tools, and display perception, while maintaining their molecular dynamics strength through CoMMBi. The trend shows a university moving from theoretical ML foundations toward large-scale computational applications and practical developer tools.

USI is expanding from core ML/computational theory into high-performance computing applications and AI-assisted software development, making them increasingly relevant for projects requiring scalable computation.

How they like to work

USI operates as both a project leader and a reliable consortium partner, with a roughly balanced split (8 coordinated vs 14 as participant). Their coordinated projects tend to be ERC grants — single-PI fundamental research — while their participant roles span large multi-partner consortia across diverse topics. With 327 unique partners across 38 countries, they maintain a broad and non-exclusive network, suggesting openness to new collaborations rather than reliance on a fixed set of partners.

USI has collaborated with 327 distinct partners across 38 countries, reflecting a wide European network with global reach. As a Swiss institution, they bridge EU and non-EU research ecosystems, which can be strategically valuable for consortium diversity.

What sets them apart

USI combines deep expertise in mathematical machine learning (geometric deep learning, graph networks) with strong computational physics (metadynamics, molecular dynamics) — an unusual pairing that few European universities offer under one roof. Their ERC track record (4 Advanced/Starting/Consolidator grants in the dataset) signals individual research excellence, while their participation in large infrastructure projects shows they can also function as team players. For consortium builders, USI offers a Swiss partner with genuine computational depth and a proven ability to both lead and contribute.

Highlights from their portfolio

• AlgoRNN
  ERC Advanced Grant (EUR 2.5M) on recurrent neural networks that learn algorithms — USI's largest single grant, running 7 years through 2024.
• CoMMBi
  ERC grant (EUR 2M) running until 2028 on computational molecular binding from atomic to cell membrane scale — their most ambitious ongoing project.
• GoodNews
  Applied geometric deep learning to fake news detection in social networks — demonstrates USI's ability to translate fundamental ML research into societally relevant applications.