Deep Learning Based Yau-Yau Algorithm for Time-Varying Filtering System
组织者
演讲者
孙泽钜
时间
2024年12月28日 21:30 至 22:00
地点
Online
摘要
Time-varying nonlinear filtering, especially with high dimensions, is a challenging issue in many application scenarios. Although the introduction of artificial intelligence greatly improves the efficiency of nonlinear filtering, its interpretability has not yet been revealed sufficiently. In this paper, we propose a deep learning based Yau-Yau algorithm framework for general time-varying nonlinear filtering problems. Two neural network structures are designed, theoretically analyzed and numerically tested for the implementation of this algorithm framework. The advantage of our proposed algorithm is that it combines both the computational efficiency of artificial intelligence and the theoretical rigor of Yau-Yau algorithm.