QxBranch’s Principal Researcher of Quantitative Analysis, Dr David Garvin, presented on innovations in quantum algorithms developed in collaboration with quantitative analysts and machine learning experts at the Commonwealth Bank of Australia in his address at the 3rd Machine Learning & AI in Quantitative Finance Conference on Thursday, March 21. These results showed the tractability of utilizing universal quantum computers and the Quantum Approximate Optimization Algorithm for portfolio optimization use cases within the bank, and broke new ground in the application of the same algorithm in the space of discrete game theory and generative adversarial networks.
Generative adversarial networks are utilized by financial organizations for building robust classifiers of data, and for generation of synthetic data where data is scarce or sensitive, such as Personally Identifiable Information (PII). Research conducted by QxBranch and the Commonwealth Bank identified new techniques for using Quantum Approximate Optimization to solve the global “social” optimum and the Nash equilibrium of a discrete congestion game, paving the way for these techniques to reproach how generative adversarial networks are optimally trained given highly discrete problem spaces. This research has broken new ground in areas identified by Perdomo-Ortiz et al.1 as potentially fruitful for the application of quantum-assisted machine learning in near-term quantum computers.
Research in this space is ongoing, and QxBranch and the Commonwealth Bank will be publishing more detailed results later in 2019.
Those interested in receiving a briefing on the potential of quantum computing to impact the financial services industry in the areas of optimization and machine learning are invited to contact us for more information.
1 Perdomo-Ortiz et al, “Opportunities and challenges for quantum assisted machine learning in near term quantum computers”, Quantum Sci. Technol. 3.3, 2018