Director of the MIPT Cognitive Modeling Center and director of the AIRI Institute laboratory Alexander Panov told how the Nobel neural network works.
“The Hopfield neural network is a computational function. It consists of several parameters that are configured without human intervention and without knowing the correct answers,” he said in an interview with Gazeta.Ru.
According to him, the Hopfield network, unlike classical ones, learns to recognize patterns automatically. During the learning process, it adjusts its parameters.
“We see certain patterns and remember them in such a way that the connection between neurons provides the least amount of energy. As a result, we save the image and can restore it if necessary,” Panov added.
Earlier it became known that the Nobel Prize in Physics in 2024 was awarded to scientists John Hopfield and Geoffrey Hinton for their discoveries that enable machine learning using artificial neural networks.
