数字技术与经济金融前沿论坛（第45期）

摘要：

Deep Reinforcement Learning (DRL) has emerged as a transformative methodology for solving sequential decision-making problems, particularly in environments characterized by uncertainty, high dimensional state spaces, and dynamic feedback. This survey provides a comprehensive overview of DRL algorithms and their applications in finance and actuarial science. We begin by reviewing foundational concepts and notable DRL algorithms, including value-based, policy-based, and actor-critic methods. We then explore their deployment in key financial domains, such as portfolio optimization, algorithmic trading, and risk management, as well as in actuarial tasks like dynamic pricing and claims reserving. The article concludes with a discussion of current challenges, research opportunities, and future directions for DRL in these domains.

主讲人介绍：

赖永增，加拿大劳瑞尔大学数学系正教授，于 1983 年和 1988 年分别在中山大学数学系获得学士学位和硕士学位，于 2000 年 1 月在美国加州克莱蒙研究生大学获得博士学位，2000 年 5 月至 2002 年 6 月在加拿大滑铁卢大学高级金融研究中心和统计与精算学系做博士后研究员，他的主要研究领域包括金融数学（衍生产品的定价与风险管理、金融计算、投资组合优化、 随机分析在金融和保险中的应用）、微分方程在金融和经济学中的应用、蒙特卡洛和拟蒙特卡洛仿真方法及应用; 机器学习及其应用, 尤其在经济金融中的应用。他在 Automatica, Computers & Operations Research, Economic Modeling, Expert Systems with Applications, Energy Economics, Finance Research Letters, Insurance: Mathematics and Economics, Journal of Computational Finance等国际期刊已经发表了 70 多篇论文。主持加拿大国家自然科学与工程基金多项，担任加拿大科学与工程国家面上基金数学统计口评审委员会委员。