Artificial neural networks, as their name suggests, are inspired by the functioning of biological neurons. These systems mimic the way neurons in the brain connect and communicate with each other to learn and solve problems. The 2024 Nobel Prize in Physics rewards John Hopfield and Geoffrey Hinton for their pioneering work in the 1980s, where they developed neural models based on physical phenomena to lay the foundations of artificial intelligence.
Artificial neural networks: at the crossroads of physics, mathematics and biology
Since that time, artificial neural networks have taken a central place in fields as diverse as voice recognition, medical diagnosis and even astrophysics. However, the implementation of these systems is energy intensive, a crucial issue at a time when the need for artificial intelligence is exploding. How could the models inspired by Hopfield and Hinton reduce the energy consumption of artificial intelligence? Answers with Julie Grollier, CNRS research director at the Albert Fert laboratory – the Thales research and development center in Palaiseau, on the campus of the Paris-Saclay polytechnic school.
