Six ETH Zurich researchers, working in the areas of artificial intelligence, robotics, biomedicine, chemistry and statistics, are being awarded prestigious Starting Grants from the European Research Council.
The Starting Grants from the European Research Council (ERC) are some of the most prestigious European funding for researchers embarking on an independent scientific career. By awarding the grants through a competitive selection process, the ERC funds particularly scientifically ambitious projects from all subject areas with up to 1.5 million euros over a period of five years. These grants are funded via the EU's Horizon Europe research programme, in which Switzerland participates financially in proportion to the benefits it derives.
ETH Zurich has been successful in the latest round of funding, with six of the around 450 grants issued across Europe going to researchers who submitted their project proposals via the university. These scientists work in the fields of biomedicine, chemistry, AI, robotics and statistics.
The award recipients and their projects:
Máté Bezdek
In chemical reactions, it is often necessary to break strong bonds between atoms or to form new ones. Chemists sometimes use helper molecules that take up an electron and a proton from another molecule or transfer them to it. This process is known as proton-coupled electron transfer (PCET). A key goal is to make this behaviour switchable, so that the same molecule can be tuned either to take up or to release a proton and an electron depending on its electrical state. In his ERC project, chemistry professor Máté Bezdek aims to develop new molecules and materials with this property, based on abundant elements such as carbon, phosphorus and sulfur. The project findings could eventually contribute to new battery materials, more sensitive chemical sensors and new catalysts.
Yuansi Chen
Yuansi Chen, a professor of mathematics, conducts research at the interface between statistics and machine learning. In his ERC project, he will investigate how uncertainties in complex data analysis can be determined when a large number of variables are unknown. The focus is on Markov Chain Monte Carlo algorithms. These computational methods are used to generate samples from complex probability distributions. This makes it possible to estimate which values for unknown variables are more or less likely based on the available data. Chen aims to make these methods faster and more reliable - particularly for challenging tasks where current approaches reach their limits. He also aims to identify which problems can be solved efficiently and how the computational methods can automatically adapt to different tasks. In this way, the project lays the foundations for more reliable, efficient and transparent data analysis.
Parth Chansoria
Parth Chansoria specialises in tissue engineering and regenerative medicine, where living cells are guided by a finely structured scaffold to form new body tissue. One potential application is the regeneration of muscle tissue after severe injury. In his ERC project, he aims to develop a new method for this purpose, based on novel soft materials for the scaffold. These materials respond to superimposed ultrasound waves: where the waves overlap, they crosslink the material, allowing to shape it into fine, precise structures. Chansoria will also develop the ultrasound device required for this process. The new materials are designed to be biocompatible, enabling the method to be used directly inside the body. The researcher plans to test the technology first in the lab, then later in animals.
Ryan Cotterell
Although well-known AI language models such as ChatGPT were originally developed for processing language, they are increasingly being used to solve logical problems and for programming. Scientists do not yet fully understand when and why these models are able to solve such tasks. Ryan Cotterell is a professor of computer science and conducts research at the intersection of artificial intelligence, linguistics and theoretical computer science. In his project, he aims to investigate the basic architecture of large AI language models, known as the transformer architecture. He wants to find out under what conditions such models can learn general rules and apply them to new cases - and when they merely utilise statistical patterns from their training data. For this fundamental research project, he intends to use mathematical methods, statistical theory and targeted experiments with such language models.
Lars Lindemann
Lars Lindemann is a professor of Algorithmic Systems Theory. In his ERC project, he investigates interactive autonomous systems such as drones or self-driving vehicles. His research focuses on how these systems can reliably perform their tasks even in the face of unforeseen events such as coordination errors, disruptions, or delays - challenges that often affect the safety of current technologies in real-world environments. Lindemann will develop new mathematical models and algorithms to describe, compute, and improve the reliability of autonomous systems. He will then test these methods in practical applications, including the coordinated use of multiple drones and ground robots for wildfire response and the management of fleets of self-driving vehicles.
Robert Baines
Robert Baines was a postdoctoral researcher in the Department of Mechanical and Process Engineering and a Branco Weiss Fellow. With his ERC Advanced Grant, he aims to develop novel robots that can change their shape, enabling them to move in a particularly versatile manner and adapt to their environment, much like living organisms. Baines left ETH Zurich at the end of August and will carry out his ERC project at the University of Oxford.
Additional project transferred to ETH Zurich
Elzbieta Gradauskaite
Elzbieta Gradauskaite conducts research at the intersection of inorganic chemistry and condensed matter physics. In her ERC project, she aims to develop a new class of materials termed multiferroic metals that combine several properties of interest for electronic applications: the materials are to be magnetic and their conductivity is to be precisely tunable by varying their chemical composition. Furthermore, they are intended to be polar in the sense that positive and negative ions within them are slightly displaced relative to one another. Using computational models, Gradauskaite will predict promising compositions and fabricate them as thin films. This approach could allow her to produce materials that would not be stable in conventional three-dimensional form. In the future, these materials could enable the development of smaller and more energy-efficient electronic and spintronic components that integrate memory and computing functions in a compact space. While originally proposed with the French National Centre for Scientific Research (CNRS) as the host institution, Gradauskaite will carry out the project at ETH Zurich, where she will take up a professorship on 1 October.
