Their groundbreaking contribution to the field of neural networks, which has transformed the way machines learn and replicate human behaviour, earned US physicist John Hopfield and British computer scientist and cognitive psychologist Geoffrey Hinton the Nobel Prize in Physics (2024). Their work, pioneering computational neuroscience and artificial intelligence (AI), has laid the foundations of modern machine learning.
Hopfield’s journey began in the 1980s, when he introduced the so-called ‘Hopfield network’, a conceptual breakthrough based on principles of physics. This type of network has the ability to store and retrieve patterns, even when the incoming information is incomplete. In that way, it emulates a form of associative memory similar to that of humans.
His proposal, based on the energy minimisation of spin systems in physics – for physicists and engineers, who will understand, spin is a property of elementary atomic particles whereby they have an intrinsic angular momentum of fixed value – allowed Hopfield’s networks to reconstruct confusing or incomplete data. In doing so, he anticipated the error correction that occurs in today’s AI systems.
Simultaneously, Hinton extended the concepts introduced by Hopfield and presented the Boltzmann machine in 1985, a model based on statistical physics that incorporated hidden layers in neural networks. A step that allowed machines to autonomously learn increasingly complex patterns. These innovations were fundamental to the later development of deep learning neural networks, which were able to classify and generate data, such as images or voices, with astonishing accuracy.
As we can see, today’s achievements in AI build on decades of interdisciplinary collaboration. After all, Hopfield and Hinton’s work was developed in fields such as physics, neuroscience and computer science. This synergy has led to technologies that revolutionise the world and shape the future, as indeed AI does.
By Manolo Barberá, senior hydraulic modeller in Amusement Logic’s Architecture Dept.
