Artificial intelligence has grown commonplace in recent years, with applications such as voice recognition, picture identification, medical diagnosis, and many more. Simultaneously, quantum technology has been shown to be capable of processing power far beyond that of even the world’s greatest supercomputer. Physicists at the University of Vienna have recently created a novel gadget called a quantum memristor, which may enable these two worlds to be combined, unleashing hitherto unimaginable powers. The experiment was carried out in partnership with the Italian National Research Council (CNR) and the Politecnico di Milano on an integrated quantum processor that operates on single photons. The findings appear in the current edition of the journal Nature Photonics.
Neural networks are mathematical models that are at the core of all artificial intelligence applications. These models are based on the biological structure of the human brain, which is comprised of linked nodes. Just as our brain learns by constantly rearranging the connections between neurons, neural networks can be mathematically trained by tuning their internal structure until they are capable of human-level tasks such as face recognition, medical image interpretation for diagnosis, and even driving our cars. Having integrated devices capable of swiftly and effectively completing the calculations necessary in neural networks has therefore become a significant research priority, both academic and industrial.
The development of the memristor in 2008 was a big game changer in the sector. The word memory-resistor, or memristor, refers to a device that adjusts its resistance based on a memory of a previous current. Almost immediately after their discovery, scientists noticed that the unusual behavior of memristors (among many other uses) was eerily similar to that of cerebral synapses. As a result, the memristor has become a critical component in neuromorphic designs.
A group of experimental physicists from the University of Vienna, the National Research Council (CNR), and the Politecnico di Milano, led by Prof. Philip Walther and Dr. Roberto Osellame, have now demonstrated that it is possible to engineer a device that has the same behavior as a memristor while acting on quantum states and capable of encoding and transmitting quantum information. A quantum memristor, in other terms. It is difficult to realize such a device since the dynamics of a memristor seem to contradict usual quantum behavior.
The scientists overcame the difficulty by employing single photons, or single quantum particles of light, and using their unique capacity to travel concurrently in a superposition of two or more routes. In this experiment, single photons are steered on a superposition of many routes as they travel via waveguides laser-written on a glass substrate. One of these pathways is used to detect the flux of photons passing through the device, and this amount regulates the transmission on the other output through a sophisticated electrical feedback mechanism, creating the required memristive behavior. In addition to demonstrating the quantum memristor, the researchers provided simulations demonstrating that optical networks with quantum memristors can be used to learn on both classical and quantum tasks, implying that the quantum memristor could be the missing link between artificial intelligence and quantum computing.
“Unlocking the full potential of quantum resources inside artificial intelligence is one of the biggest challenges of present research in quantum physics and computer science,” says Michele Spagnolo, the journal’s lead author. The University of Vienna’s Philip Walther group has also recently proved that employing quantum resources and borrowing strategies from quantum computing allows robots to learn quicker. This remarkable accomplishment signifies another step toward a world in which quantum artificial intelligence is a reality.