Explorations in Quantum Machine Learning
PI: Vasant Honavar (IST)
PI will be responsible for tuition support.
Proposal description:
Quantum machine learning (QML) aims to leverage quantum computers with to surpass the performance of classical machine learning (CML) implemented on classical computers. The PI is an expert in machine learning, but a novice in quantum computation. Hence, this project will be carried out in collaboration with colleagues at Penn State and DOE National Labs who have expertise in quantum computing. The goal of this exploratory project are two-fold: (1) To rigorously empirically compare a broad class of QML algorithms with their CML counterparts (2) Apply QML methods that demonstrate clear advantage over their CML counterparts to challenging real-world problems that the PI has extensive prior experience, including prediction of protein-protein and protein-RNA interfaces, interactions, and complexes (in collaboration with structural biologists that the PI has been working with for over a decade); learning of universal interatomic potentials and universal fundamental thermodynamic equations (in collaboration with computational material scientists that the PI has been working with over the past four years).
Specific aims of the research include:
• Rigorous empirical comparison of QML algorithms with their CML counterparts on a broad range of machine learning benchmarks
• Apply state-of-the-art QML algorithms to prediction of protein-protein and protein-RNA interfaces, interactions, and complexes
• Apply state-of-the-art QML algorithms to train universal interatomic potentials (from large density functional theory data) and universal fundamental thermodynamic equations
The long-term goals of this research are to develop innovative QML algorithms that offer substantial advantages over their CML counterparts for a broad range of real-world applications. This project lays the groundwork by allowing the PI and his team to develop the necessary experience with QML, gather preliminary data to support competitive collaborative QML research proposals
Connection to ICDS Mission: This project directly supports the ICDS mission by helping expand research expertise and capacity as it relates to quantum machine learning. It is aligned with the ICDS research priorities in quantum computing, AI, and Data Science.
Ideal student background: Good knowledge of machine learning and familiarity with quantum computing (ideally quantum machine learning).