Scientists say that they made a breakthrough after developing a quantum computer technique to execute automatic learning algorithms that surpass classic cutting -edge computers.
The researchers revealed their results in a study published on June 2 in the journal Nature photonics.
Scientists have used a method based on a quantum photonic circuit and a tailor -made automatic learning algorithm.
Using only two photons, the team technique has successfully demonstrated an increased speed, accuracy and efficiency compared to standard classic calculation methods to execute automatic learning algorithms.
Scientists say it is one of the first times that quantum automatic learning has been used for real world problems and offers advantages that cannot be simulated using binary computers. In addition, due to its new architecture, it could be applied to quantum computer systems with only one qubit, they said.
Unlike many existing methods to achieve acceleration thanks to quantum hybrid-classic calculation techniques, this new method does not require tangle Gates. Instead, it is based on the injection of photons.
In relation: “ Science is resolved ”: IBM to build a quantum computer monster of 10,000 qubit by 2029
Essentially, the team used a femtosecond laser – a laser that emits light in extremely short pulses measured in femtoseconds (10⁻¹⁵ seconds) to write on a borosilicate glass substrate to classify data points from a set of data. The photons were then injected into six distinct configurations, which were processed by a hybrid quantum system.
Scientists have determined where photonic measures surpassed those conducted by classic calculation by measuring how long the photons took to finish the quantum circuit. They then isolated the processes where quantum treatment provided advantages and compared the results with conventional outings.
The researchers found that the experiments carried out using the photonic quantum circuit were faster, more precise and more energy efficient than those conducted using only conventional calculation techniques. This stimulated performance applies to a special automatic learning class called “Automatic learning based on the kernel” which can have a myriad of applications through data sorting.
While deep neural networks have become An increasingly popular alternative to nucleus methods For automatic learning over the past decade, the kernel -based systems have seen a resurgence In recent years, due to their relative simplicity and their advantages during their work with small data sets.
The team experience could lead to more effective algorithms in the fields of natural language treatment and other supervised learning models.
Perhaps even more important, the study presents a new method to identify the tasks that quantum computers excel in hybrid computer systems.
Researchers say that the techniques used are evolving, which means that they could lead to even better performance as the number of photons or qubits increases. This could, in turn, make it possible to develop automatic learning systems capable of exceeding the limits of today’s models, which are more and more confronted Energy consumption limitations Due to the massive energy needs necessary to process data via electronics.
Researchers say that their techniques “will open the door to hybrid methods in which photonic processors are used to improve the performance of standard automatic learning methods”.