Quantum Machine Learning for Financial Data Streams

金融数据流的量子机器学习

• 批准号: 10073285
• 金额: $ 42.85万
• 依托单位: STANDARD CHARTERED BANK
• 依托单位国家: 英国
• 项目类别: Feasibility Studies
• 财政年份: 2024
• 资助国家: 英国
• 起止时间: 2024 至 无数据
• 项目状态: 未结题
• 来源: https://gtr.ukri.org/projects?ref=10073285
• 关键词: Quantum; Machine; Learning; Financial; Data

**The Opportunity:** Financial institutions need to continuously interpret complex data streams to extract information necessary for providing accurate credit risk evaluation, managing market-making services, and predicting emissions in the context of green finance. Current classical machine learning (ML) techniques used to assist and provide insights to these services have limitations as these data streams evolve in complexity. There are three key challenges that financial institutions are seeking to address in an effort to improve their offerings: (1) Providing clients accurate credit-risk evaluation services, (2) Offering competitive rates for market-making services, and (3) Predicting emissions for informed sustainable finance decisions in line with ESG targets. Improving upon the current classical ML approaches could result in reduced risk, better market rates and targeted sustainable investments for financial institutions and their customers.**The Approach:** Recent quantum computing advances have the potential to offer significant improvements to the computations financial institutions rely on to improve upon efficiency, to reduce risk, to provide better service to customers and to develop personalised products. The team's offering using cutting-edge quantum machine learning techniques, running on an optimised full-stack Rigetti platform, will offer financial institutions a vertically integrated solution, allowing them to use the full capability of NISQ-era quantum computing. We will develop quantum signature kernels and leverage the results to enhance Rigetti's recent breakthroughs in quantum kernels. We will benchmark the results against classical ML methods for streamed data. Additionally, we will build and study quantum algorithms for computing efficient signatures and their inner products for long and high-dimensional data streams. **Innovation and Benefits:** A successful project outcome will have significant benefits for the UK financial sector and the quantum computing industry, including the participating organisations. Accelerating the development of quantum machine learning for financial data streams will enable Standard Chartered to be an industry leader in a future quantum-ready economy and continue to provide the best possible services to its clients. Developing quantum-enabled solutions will also bolster the UK finance sector. Rigetti will be able to accelerate its work to achieve narrow quantum advantage, the point at which a quantum computer outperforms the best classical resources. The project will also benefit Imperial College London by providing a framework for and use cases to test new quantum machine learning tools. Making these tools open access will further allow UK academics to test state-of-the-art quantum algorithms for their own applications (possibly beyond those in this proposal).

**机会：**金融机构需要不断解释复杂的数据流，以提取提供准确的信用风险评估，管理营销服务以及在绿色金融背景下预测排放所必需的信息。当前的经典机器学习（ML）技术用于协助和为这些服务提供见解具有局限性，因为这些数据流的复杂性不断发展。金融机构寻求解决的三个关键挑战以改善其产品：（1）为客户提供准确的信用危险评估服务，（2）为市政服务提供有竞争力的价格，以及（3）预测与ESG目标一致的知情可持续融资决策的排放。改善当前的经典ML方法可能会导致风险降低，更高的市场利率和针对金融机构及其客户的可持续投资。在优化的全栈Rigetti平台上运行的最先进的量子机学习技术，该团队的产品将为金融机构提供垂直集成的解决方案，从而使他们能够使用NISQ时代量子计算的全部功能。我们将开发量子签名内核，并利用结果来增强Rigetti最近在量子内核中的突破。我们将针对流媒体数据的经典ML方法进行基准测试结果。此外，我们将在长期和高维数据流中构建和研究用于计算有效签名及其内部产品的量子算法。 **创新和收益：**成功的项目成果将为英国金融部门和量子计算行业（包括参与组织）带来重大好处。加速为财务数据流的量子机学习的开发将使宪章成为未来量子准备经济的行业领导者，并继续为其客户提供最佳服务。开发支持量子的解决方案还将加强英国财务部门。 Rigetti将能够加速其工作以实现狭窄的量子优势，量子计算机优于最佳经典资源的点。该项目还将通过为测试新的量子机学习工具的框架和用例提供框架和用例，从而使伦敦帝国学院受益。使这些工具开放访问将进一步使英国学者可以为自己的应用测试最先进的量子算法（可能超出了本提案中的量子）。