Using quantum computers to speed up time series analysis
Using quantum computers to speed up time series analysis lead image
An essential task in time series analysis is the extraction of physically meaningful information from noisy time series data. One algorithm employed for this purpose is singular spectrum decomposition (SSD), used in applications such as sleep apnea detection from an unprocessed ECG, processing and analysis of brain waves, and gravitational wave analysis.
The main limitation when scaling up the SSD algorithm for larger data sets is the singular value decomposition (SVD) calculation, which is required for each subroutine. Postema et al. used a quantum computer to implement a new method to make this process significantly faster.
The researchers aimed to convert the standard classical SSD algorithm into a variational hybrid quantum-classical algorithm called quantum singular spectrum decomposition (QSSD). They wanted to determine whether quantum computing offered freedom from the scaling limitations inherent to classical SVD.
“We have shown that performing an SVD on a quantum computer is possible in two different ways, digital and analog, the latter of which has never been proven before,” said author Jasper Postema. “This approach provides a native way of performing variational optimization for neutral atom quantum devices.”
The researchers also showcased that, while quantum SVD performs poorly in isolation, it is stable against quantum errors when combined with classical SSD. However, existing technology is not enough to provide a significant speed increase.
“This is simply a first step, and eventually quantum SVD will be modified, or new ideas will be used, that will render such algorithms viable,” said Postema. “Most likely, an efficient algorithm would require fault tolerant quantum resources and efficient techniques for the encoding of classical data that are yet to be discovered.”
Source: “Hybrid quantum singular spectrum decomposition for time series analysis,” by J. J. Postema, P. Bonizzi, G. Koekoek, R. L. Westra, and S. J. J. M. F. Kokkelmans, AVS Quantum Science (2023). The article can be accessed at https://doi.org/10.1116/5.0139846 .
