Home > News & Events > Events Content
Speaker:Ning Ning, Dctor, Department of Statistics and Applied Probability, UCSB
Date:July. 19, 2022
Time:09:30-11:30
Location:Tencent Meeting,ID:873 393 528
Sponsor:Research Center for Mathematics and Interdisciplinary Sciences
Research Center for Nonlinear Expectation
Abstract:
Large financial systems are typically modeled by interacting particle systems (IPS) that are nonlinear stochastic processes with strongly interrelated components. Both the joint probability distribution and the marginal probability distributions are intractable, except very simple cases in very low dimensions (usually dimension being two). Calibration of the parameters of such a complicated model is indeed an extremely difficult problem requiring evaluations at many points in the parameter space to optimize (Carmona et al. (2009), Finance and Stochastics). This great methodological challenge has been conquered very recently in Ning and Ionides (2021a) through the iterated block particle filter algorithm, for learning high-dimensional parameters over partially observed, nonlinear, and interacted time series models on a general graph, which outperforms the iterated ensemble Kalman filter algorithm (Li et al. (2020), Science) and the iterated filtering algorithm (Ionides et al. (2015), PNAS) in corresponding scenarios. There are open and natural questions such as, how to practically learn high-dimensional parameters over complicated financial models which are usually continuous-time models with their unique properties (such as extra mutual correlations in driving Wiener processes), and especially how to incorporate and complement existing well-functioning algorithms. In this talk, I will generalize a well-established financial IPS to a general graphical setting, demonstrate methodologies on merging rigorous machine learning algorithms into financial paradigms, conduct high-dimensional volatility learning over a large portfolio of assets even covering private equities that have no market prices, and shed light on more accurate control and policymaking over large financial systems.
For more information, please visit:
https://www.view.sdu.edu.cn/info/1020/167631.htm