Los Alamos National Laboratory Research Paves Way for Simplified Quantum Machine Learning

New Los Alamos National Laboratory research found that a quantum neural network can be trained using a small amount of simple data, paving the way for less complex machine learning on quantum computers. Authors of the paper include researchers from LANL, the United Kingdom, Switzerland and Germany.

Existing quantum computers use long circuits that produce error-causing noise, preventing the systems from employing their full processing capabilities. Using simple data creates less complicated circuits that are easy to implement and less noisy, allowing quantum computers to perform practical work, including completing a computation, LANL said.

The research team aims to develop more efficient algorithms for quantum ML. Zoe Holmes, professor of physics at Ecole Polytechnique Federale de Lausanne and a co-author of the paper, explained that quantum ML at simple states is easier to prepare and run on near-term quantum computers. 

The study also found that classical computers can be used to compile quantum algorithms, a capability that allows programmers to reserve quantum computing resources for tasks that traditional computers cannot support.