The Italian CNR has developed a unique algorithm of its kind, capable of greatly improving learning a quantum computer. Enough to beat a similar study prepared by Google. The deep learning developed by Italian researchers allows a great step forward for the future of quantum computing.
The Italian quantum computer “learns” better than Google’s
A study published in Nature, in the Communication Physics insert, shows how the CNR deep learning algorithms for quantum computers are exceptional. Even better than those of a patent filed by Google in solving the “logic gates problem”. Lorenzo Moro of the CNR explains it with a comparison with playing cards: “We have developed with deep learning an algorithm capable of finding the right order to play five or six cards available to the computer, even with sequences of hundreds of plays long, choosing one by one the right ones to form the entire sequence “.
The “trump card” compared to the Google project lies in the possibility of use logic gates automatically, while Google had to start training all over again each time. This can take deep learning, the continuous learning of artificial intelligence, to another level.
Companies like Google had accelerated this learning but not in a revolutionary way as much as the study of the CNR. Enrico Prati, the researcher at the head of the group that carried out the research, explains: “The technology of the quantum computer allows to solve computing problems that are impossible with ordinary tools and, in this scenario, quantum compilers become fundamental for an efficient control of systems. “.
The research conducted with the collaboration of Matteo Paris of the State University of Milan e Marcello Restelli of the Politecnico di Milano, now has a patent that can also be used by other researchers. A program that manages to solve in few milliseconds the problem of which logic gates use. The CNR polyvalent algorithm can be applied to any computer with quantum logic gates. The applications are many. And all to be discovered: the great work of the CNR researchers is only the beginning of this computational revolution.