Scientists have improved quantum artificial intellectuals Wolf | 14.07.2020
Scientists of the leading technology institutes in Germany, Korea and South Africa have perfected a machine learning technique for use in quantum computers.
The team from the Korean Institute Korea Advanced Institute of Science and Technology (KAIST), in collaboration with researchers from the University of KwaZulu-Natal (UKZN) in South Africa and Cybernetics Data
in Germany has introduced quantum classifiers with the “adapted quantum nuclei”. According to the statement of experts, the input data in accordance with a new principle of classification is presented using the
map of quantum characteristics or quantum internal data and the classification based on kernel function that measures the closeness of test data to training data. The ability of quantum
computers to efficiently manipulate the data in the space of quantum characteristics opens up completely new possibilities for quantum machine learning algorithms.
Quantum machine learning promises to be one of the must-have applications for quantum computing. In machine learning one of the fundamental problems is the classification task required
to recognize patterns in the labeled training data to assign a label to new, previously unseen data. First quantum artificial intelligence used classic
approaches that do not allow to use the full potential of quantum technologies.
The principles, formulated by scientists from Germany, Korea and Africa, according to ScienceDaily steel
an important step towards building quantum algorithms computing.
Dr. Daniel Park from KAIST, one of the lead authors of the study, said that the possibility of a systematic adaptation of the “quantum core” makes it a good candidate for holding
a wide range of real research.
In addition to the “quantum core” team has also developed the principle of “quantum branching” method which allows you to start the training Protocol from scratch, even when all the marked test and
the reference data independently from each other are encoded in a separate qubits.
To demonstrate the usefulness of a new principle of classification, Karsten the Form of Data Cybernetics tested the classifier on quantum quantum IBM computer with five qubits
which is freely available to public users via cloud services. Results Explorer details not described, but called “promising.”
Journal: Journal IT-News, Subscription to magazines