The Swedish Academy of Sciences has announced the scientists who are honored to receive the Nobel Prize in Physics. This year they were John Hopfield and Geoffrey Hinton. Their developments are devoted to artificial neural networks.
Nobel for neural networks
On October 8, the Royal Swedish Academy of Sciences announced that this year’s Nobel Prize in Physics would be awarded to John Hopfield and Geoffrey Hinton. There is nothing surprising in the fact that two scientists will receive this prestigious award at once. What is much more interesting is what they are awarded for.
In the media, the subject of the work for which Hopfield and Hinton won the Nobel Prize is succinctly characterized as “for neural networks.” And given the boom of programs that draw pictures and write poems, it seems like a trend following because, at first glance, this topic is primarily about programming. In other words, although very complex and involving calculations, it has nothing to do with pure physics.
Actually, neither Hopfield nor Hinton were involved in creating the code for ChatGPT. The subject of their case concerns quite different neural networks, and they didn’t deal with them directly either, but made two separate discoveries without which these systems would have been possible.
Accomplishments of scientists
John Hopfield created associative memory that can store and retrieve images and other types of data structure. His discovery is based on the use of such a characteristic of elementary particles as spin. Based on it, he created the principle of how an artificial neural network can be caused to “remember” that it has already seen something like this, similar to the way a human does.
Geoffrey Hinton took a neural network built on Hopfield’s discovery and created something called a “Boltzmann machine” based on it in 1985. It was the first system capable of detecting certain elements in a data set.
In order to understand the significance of Hopfield and Hinton’s work, it is necessary to realize that concerning what we now call neural networks, it would be correct to specify “artificial”. In fact, they are an attempt to reproduce in non-organic systems the principles by which we think our own central nervous system works.
According to x.com/NobelPrize
