Skip to content
Technical University of Munich

News Department of Chemistry

Science advancements powered by machine learning

AMC, ORIGINS, Chemistry, Physics |

Nobel Prize winners used principles from physics to develop methods that form the basis for today's powerful machine learning. How are researchers at TUM using these fundamental discoveries today?
Interior view of Leibniz Supercomputing Centre on Campus Garching. Picture: Andreas Heddergott / TUM

The Nobel Committee announced last week that the 2024 Physics Prize has been awarded to John Hopfield and Geoffrey Hinton for machine learning (ML) and artificial neural networks. While computers cannot think as humans do, machines can now simulate such functions as memory and learning. The 2024 physics Nobel laureates have facilitated this shift using fundamental concepts and methods from physics.

In the TUM School of Natural Sciences, we asked our professors to reflect on the impact of machine learning and deep learning/artificial neural networks on their current research. 

Machines identify, sift, calculate and make better use of scientific data

In research, we produce a lot of data, information that requires time and expert insights to be useful. The prize awarded to Hopfield and Hinton highlights how these methods can help sort and analyze the huge amounts of data produced in experimental and theoretical research.

“I’m convinced that by applying AI to some of the most demanding physics environments we can advance both AI and physics at the same time,” says Prof. Lukas Heinrich, TUM Chair of Data Science in Physics. At the announcement last week in Sweden, the Large Hadron Collider was an example of how the laureates’ works have contributed to scientific research since the 1980s, “It has deeply changed the way we do our work. While before, we were limited by our own imaginations in how to extract insights from the data at the Large Hadron Collider, we can now exploit AI to consider even the subtlest patterns that otherwise would be inaccessible,” Heinrich reports.

Prof. Helge Stein, TUM Professor of Digital Catalysis, agrees that the impact of the Physics Nobel Laureates reaches beyond the traditional physics boundary with positive consequences for chemistry, “While theoretical chemists have been using machine learning methods much earlier than some experimentalists, the impact on materials chemistry has been tremendous.” He remembers his first ML publications around 2013 when many critics exclaimed, “this is just a phase!”. The 2024 Nobel Prize validates the contribution of ML to advancing and accelerating the fields of natural sciences.

Prof. Patrick Rinke, TUM Chair of AI-based Materials Science, agrees that the impact of machine learning and deep learning has been huge: “No other development has had quite the same effects because machine learning can be and is used in all of physics and all of chemistry.” Rinke continues, “In every sub-discipline, machine learning has led to new insights, facilitated acceleration of data generation and understanding, and has enabled research that simply wasn't possible in this form before.”  In his own research, ML has allowed Rinke to tackle a much wider range of scientific problems. “My previous expertise was in atomistic modeling. With machine learning, we can go beyond the atomic scale and address challenges with societal impact, like sustainable materials development (such as organic electronics).”

Nobel Prize press release

Finding patterns of information “They trained artificial neural networks using physics” October 08, 2024. https://www.nobelprize.org/prizes/physics/2024/press-release/ 

 

More information and links

 

Press contact

communications@nat.tum.de 

◄ Back to: Chemistry
To top