A Quantum Annealer for Subset Feature Selection and the Classification of Hyperspectral Images

The project uses hyperspectral images (HSIs) to benchmark the D-Wave Advantage Quantum Annealer for classifying Earth Observation data and to identify future challenges when analyzing real-world datasets. HSIs showing objects belonging to several distinct target classes are characterized by dozens of spectral bands being available. However, some of these spectral bands are redundant and/or noisy, and hence, selecting highly informative and trustworthy bands for each class is a vital step for classification and for saving internal storage space. The selection problem of highly informative bands is formulated as such that it can be solved by both a quantum annealer (QA) and a conventional machine. In a second step, binary classifiers were implemented on the QA and on a conventional system to classify a real-world dataset in Earth observation - the well-known AVIRIS HSI of Indian Pine, north-western Indiana, USA. It was found that the QA selects the informative bands correctly and generates correct classifications of similar quality compared to conventional approaches. However, it must be noted that the dataset used here was a prerequisite for the QA to perform well and that the study aimed to find a proper dataset in Earth observation to benchmark existing quantum algorithms.

References: doi: 10.1109/JSTARS.2021.3095377.