In 1944 a work by the great physicistphysicist Erwin SchrödingerErwin Schrödinger whose title was “ What is life? “. His reflections based in particular on what we call statistical mechanics would greatly influence the discoverers of the structure ofADNADN, Francis CrickFrancis Crick and James Watson. The year before, Warren McCulloch and Walter Pitts had begun to lay one of the foundations of the revolution ofartificial intelligenceartificial intelligence (IA) era XXIe century, in the form of networks of neuronesneurones artificial models inspired by the behavior of biological neurons and Deep Learning, which some wonder if it does not contain stretchingstretching access to artificial consciousness.
These are some of the other pioneers of artificial intelligence with neural networks that the Royal Swedish Academy of Sciences has just rewarded today with a Nobel Prize in physics for the year 2024. These are the physicist states -unian John Joseph Hopfield, born July 15, 1933, and the British-Canadian computer scientist and cognitive psychologist Geoffrey EverestEverest Hinton, born December 6, 1947.
An artificial intelligence or AI can do many things, but how does an AI actually work? Discover through the example of the strategy game, Go, how artificial intelligence can be programmed to play this game with billions of different configurations. Machine learning, deep learning and reinforcement learning will no longer hold any secrets for you! A video animation co-produced with The Witch Spirit. © CEA
AI with deep learningthat is to say thedeep learningdeep learning with artificial neural networks, had resumed its growth a few years ago thanks to the Frenchman Yann LeCun as well as the Canadians Yoshua Bengio and already Geoffrey Hinton, who received the 2018 Turing Prize on March 27, 2019, the equivalent of the Nobel in computer science, awarded by theAssociation for Computing Machinery (ACM).
The three researchers, relying on the work of John Hopfield which also proved crucial, had blown up a locklock dating back to the late 1960s when Marvin Minsky and Seymour Papert believed they had shown that too much hope had been placed in the learning possibilities of AI based on the work of Frank Rosenblatt in 1957 with his famous Perceptron.
In this video, Mireille Fares explains what an artificial neural network is and how it works. This video was produced as part of the Science Festival 100% digital 2020. © Sorbonne University
The Nobel Prize committee today recognizes the importance of the paradigm shift made by John Hopfield in the early 1980s not only with what is now known as Hopfield neural networks (Hopfield network))but also by showing that the behavior of these new networks could be compared to that of what physicists call the Ising model. It is a theory from statistical mechanics describing the thermodynamic and magnetic properties of matter in the form of a population of particles endowed with an angular momentum and a magnetic moment of its own like those of electronselectrons.
Geoffrey Hinton would make a new connection with another area of statistical mechanics, namely the kinetic theory of gazgaz which we owe to Ludwig Boltzmann and which in the hands of Hinton gave the Boltzmann machinesee on this subject the book, free online, by the late David MacKay who was one of Hopfield’s doctoral students and of course the press releases from the Nobel Foundation for the general public and those who are educated in physics and computer science .
The revolution of machine learning who gave birth to self-driving cars, facial recognition and learning robots dates back to John Hopfield, whose career is as fascinating as the technologies his ideas helped develop. To obtain a fairly accurate French translation, click on the white rectangle at the bottom right. English subtitles should then appear. Then click on the nut to the right of the rectangle, then on “Subtitles” and finally on “Automatically translate”. Choose “French”. © The Franklin Institute
Recall that in recent years, Yoshua Bengio and Geoffrey Hinton have expressed their concern with the rise of deep learningfearing a possible scenario at the Terminator in a much closer horizon than he previously thought.
There is no guarantee of security in the face of advances in artificial intelligence, warns AI pioneer Geoffrey Hinton. He shares his thoughts on the benefits and dangers of AI with Scott Pelley. To obtain a fairly accurate French translation, click on the white rectangle at the bottom right. English subtitles should then appear. Then click on the nut to the right of the rectangle, then on “Subtitles” and finally on “Automatically translate”. Choose “French”. © 60 Minutes
Related News :