主 讲 人: 南京理工大学, 丁凯琳 讲师
报告时间:2025年7月7日 下午15:30-16:30
报告地点:览秀楼105学术报告厅
报告摘要: In this paper, we derive novel non-asymptotic $L_1$ and $L_2$ error bounds for kernel estimators of the density and its derivatives from data. The error bounds are explicit functions of the bandwidth, which allow us to determine the optimal bandwidth by minimizing these non-asymptotic error bounds. Assuming a general kernel function, the optimal bandwidth can be determined through solving two algebraic equations. Extensive numerical experiments demonstrate that the proposed fully automatic data-driven bandwidth selection method compares favorably with existing literature.
主讲人简介:丁凯琳,理学博士,南京理工大学经济管理学院应用经济系讲师。南开大学理学博士,美国伊利诺伊大学香槟分校联合培养博士,中科院数学与系统科学研究院 管理科学与工程博士后。主要研究方向有金融衍生品定价、金融风险管理、随机仿真等。在SSCI和SCI期刊Journal of Futures Markets、Quantitative Finance、ACM Transactions on Modeling and Computer Simulation等期刊发表论文数篇。