Regarding the recently announced Nobel Prize in Physics awarded to John Hopfield and Geoffrey Hinton, Mário Figueiredo, our researcher and professor of Técnico, from the field of learning machines, has shared his thoughts in Público online today.

"In 1982, John Hopfield gained recognition for applying statistical physics models, originally used to study magnetic properties in materials like spin glasses, to propose a class of virtual neural networks that later became known as Hopfield networks. This work opened up new pathways for understanding information processing in both the brain and machine learning systems."

"Geoffrey Hinton, in 1985, together with Terrence Sejnowski (a former student of Hopfield), extended the application of statistical physics even further. They introduced what are now known as Boltzmann machines, named after Ludwig Boltzmann, the 19th-century physicist regarded as the 'father' of statistical physics."

In Mário Figueiredo's view, while Hopfield and Hinton's contributions are celebrated, they were not the first to bridge physics and machine learning. Notably, Shun'ichi Amari, a pioneering Japanese researcher, made significant advances in this field nearly a decade before Hopfield, with ideas that largely parallel the development of Hopfield networks.

Nevertheless, he says, this Nobel recognition underscores the power of interdisciplinary thinking and highlights the profound influence that physics can have on seemingly unrelated fields. It serves as an invitation for machine learning and AI researchers to embrace a physicist's mindset.

Read the full article here: