Compressing Earth science datasets with quantum-assisted machine learning algorithms

Quantum computing has the potential to improve performance, decrease computational costs and solve previously intractable problems by exploiting quantum phenomena such as superposition, entanglement, and tunneling. One area where tools are being built with near-term quantum devices is machine learning. As a result of increased computational power, advances in deep learning and availability of data, machine learning is fundamentally changing how we interpret data as researchers. We describe two hybrid quantum-classical machine learning algorithms quantum-assisted associative adversarial network and the quantum variational autoencoder and their application to generating and compressing data from NASA Earth Exchange (NEX) GIMMS dataset. Future improvements in these application areas may translate to a reduction in cost and improve the quality and quantity of metadata available for researchers. Though there has been significant progress made in the last decade, the quantum hardware is still rudimentary. Implementations of these algorithms do not compete with state-of-the-art classical implementations. However, it is essential to demonstrate hybrid architectures on problems the Earth Sciences community face, to lay the foundation for future use cases and inform quantum-based computational hardware development.