John Hopfield and Jeffrey Hinton have been awarded the Nobel Prize in Physics “for their seminal discoveries and inventions that enable machine learning using artificial neural networks.” The laureates will share the prize of 11 million Swedish kronor (over a million dollars). This was reported by the Nobel Committee.
The Nobel Committee’s press service said this year’s winners of the Nobel Prize in Physics are a 76-year-old British scientist and a 91-year-old American scientist Hopfield. They used physics tools to develop methods that are the basis of modern, powerful machine learning.
John Hopfield created an associative memory that can store and reconstruct images and other types of data patterns. Geoffrey Hinton invented a method that can autonomously find properties in data, thus performing various tasks, such as identifying specific elements in images.
“When discussing artificial intelligence, we often refer to machine learning using artificial neural networks. This technology was originally inspired by the structure of the brain. In an artificial neural network, brain neurons are represented by nodes with different values. These nodes influence each other through connections, which can be likened to synapses and made stronger or weaker. The network learns by developing stronger connections between nodes with simultaneously high values. This year’s honorees have been doing important work with artificial neural networks since the 1980s,” the press release said.
John Hopfield created a network that stores and recreates patterns. The nodes of this network are pixels. Hopfield’s network is based on physical laws that describe the properties of materials due to the atomic spin system. This phenomenon makes each atom a tiny magnet.
The entire network is described similarly to the energy of the spin system in physics, and the network is trained to find a value for the connections between nodes so that the stored images have a low energy level.
When a distorted or incomplete image is fed into the Hopfield network, it gradually processes the nodes. It updates its values to reduce the system’s energy consumption. This way, the network gradually finds the stored image most similar to the distorted one.
Geoffrey Hinton used the Hopfield network to create a new network that uses a different method, the Boltzmann machine. This machine can learn to recognize characteristic elements in a specific data type.
Hinton used tools from statistical physics, a science that studies systems with many similar components. The machine learns from examples that are likely to occur during its operation.
The Boltzmann machine can classify images or create new patterns based on the patterns it has been trained on. Hinton continued to develop this technology, which was the impetus for the rapid development of machine learning.
“The work of the laureates has already brought great benefits. In physics, we use artificial neural networks in a wide range of areas, for example, to develop new materials with specific properties,” said Ellen Moons, chair of the Nobel Committee for Physics.
