Sign In

Communications of the ACM

ACM TechNews

Quantum Machines Learn 'Quantum Data'


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
A quantum computer during assembly.

Quantum-enhanced machine learning can be applied to quantum data to surmount a critical bottleneck, according to researchers at Russia's Skolkovo Institute of Science and Technology.

Credit: Justin Fantl/Rigetti Computing

Scientists at Russia's Skolkovo Institute of Science and Technology (Skoltech) have demonstrated that quantum-enhanced machine learning can be applied to quantum data, surmounting a critical bottleneck.

When applying quantum algorithms to classical data, the information must be stored or otherwise represented by a quantum processor before quantum effects can be employed; this data-readin problem limits the possible acceleration of quantum-enhanced machine learning algorithms.

The Skoltech team combined quantum-enhanced machine learning with quantum-enhanced simulation to explore phase transitions in many-body quantum magnetic problems, in effect training quantum neural networks using only quantum states as data.

This bypasses the data-readin problem by feeding in quantum mechanical states of matter.

Skoltech's Alexey Uvarov said the study is "a step towards understanding the power of quantum devices for machine learning."

From Skolkovo Institute of Science and Technology (Russia)
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found