German physicists have developed a quantum autoencoder to clean data from noise. Using the constructed neural network, the researchers have successfully demonstrated the purification malacorange salesapprole state. Work published in the journal Physical Review Letters.
Machine learning is a powerful tool for working with data, especially when data becomes much. Some machine learning technique allow you to reveal important patterns in the data. One of the popular methods are neural networks autoencoder. For example, they were used to highlight the birds singing in the forest.
At the same time, quantum computing, operating fragile entangled quantum States can outperform classical computation in certain tasks, such as modeling of physical systems or the factorization of huge numbers. More about the possibilities of quantum computing, read our article “Quantum alphabet”. However, today’s quantum computers are quite noisy, and this noise spoils the confusion that leads to a strong decrease of the quantum efficiency of the devices. One promising solution is the application of techniques of machine learning to quantum computing for the purpose of cleaning quantum data from the noise. Learn more about it is told in mini-lectures of the physicist Sergei Filippov in PostNuke:
Physicists Dmytro Bondarenko (Bondarenko Dmytro) and Polina Feldman (Naomi Feldmann) from the University of Leibniz, proposed to use autoencoder to silesaurus quantum States in order to save confusion, even at high noise levels. To do this, scientists have proposed to associate to each qubit neuron in autoencoder with a direct link, the architecture of which allows to identify the contribution of noise in the quantum state of qubits.