讲座题目：Intelligent Annuities: Adaptive Retirement Planning with Deep RL（智能年金：基于深度强化学习的自适应退休规划）

主讲人：金卓 澳大利亚麦考瑞大学商学院 教授

讲座时间：2025年12月15日14:00

讲座地点：学院229

讲座主题摘要：

Abstract: This talk introduces a deep reinforcement learning framework specifically designed for the annuity market. We present an approach that dynamically optimizes portfolios combining variable annuities and traditional assets by directly learning from interactions with a simulated environment. The model integrates key annuity market uncertainties—like longevity risk and product-specific guarantees—into its decision process. By leveraging neural networks, the algorithm learns adaptive strategies to allocate between annuities and financial assets in response to evolving personal and market conditions. Our results demonstrate that this method provides a robust, data-driven, and personalized strategy for managing retirement income in complex annuity markets.

本报告介绍了一个专为年金市场设计的深度强化学习框架。研究者提出一种方法，通过直接从与模拟环境的交互中学习，动态优化包含变额年金与传统资产的投资组合。该模型将年金市场的核心不确定性—— 如长寿风险及产品特定担保条款 —— 融入决策过程。借助神经网络，该算法能够学习自适应策略，根据个人状况与市场环境的动态变化，在年金与金融资产之间进行配置。研究结果表明，该方法在复杂的年金市场中，为退休收入管理提供了一套稳健、数据驱动且个性化的解决方案。

主讲人学术简介：

金卓，澳大利亚麦考瑞大学(Macquarie University)商学院教授，研究方向包括最优控制论在精算中的应用，数理金融，金融科技，机器学习与金融交叉。在Insurance Mathematics and Economics, European Journal of Operational Research, Journal of Risk and Insurance, SIAM Journal on Control and Optimization, Automatica等期刊发表论文60余篇。他是北美精算学会（Society of Actuaries）成员。