Improving Machine Learning via Nested Quantum Annealing Correction

Unsupervised machine learning, where algorithms seek to extract patterns from data without predefined labels, is a growing area of interest with application to many different fields. However, the training of such algorithms is difficult, due to the intractability of traditional sampling techniques. Quantum annealing, which has the potential to sample more efficiently and from a wider range of distributions than classical methods, may help lead to future advances in the development of unsurpervised learning algorithms. However, the effective temperature at which experimental quantum annealers sample scales in general with the size of the problems. As such, error correction schemes are needed to counter this trend if there is to be some benefit in using quantm annealing. To this end, we have applied nested quantum annealing correction (NQAC), a scalable and generalizable error correction scheme, to a reduced MNIST dataset, which consists of black-and-white images of hand-written integers, and tested whether NQAC's ability to provide an effective temperature reduction leads to an increase in learning performance with nesting level.