Wenhan Li
Academic and research departments
Computer Science Research Centre, School of Computer Science and Electronic Engineering, Faculty of Engineering and Physical Sciences.About
My research project
Statistical machine learning for sequential Monte Carlo methods- Regime-switching differentiable particle filters
In real-world applications, both dynamic and measurement models in state-spaces models (SSMs) can be constituted by a variety of modes and be switchable. For instance, in object tracking, targets can exhibit diverse status, e.g., people may stroll gently, ride in heavy traffic, or cruise on motorways at different times. On the other hand, the observational measurement can be susceptible to various conditions such as obstacle sheltering or weather evolving. Therefore, it leads to an additional complexity layer of the regime mixture uncertainty in SSMs, i.e., regime-switching state-space models. This project leads to analyse such complicated dynamical systems with satisfied flexibility and scalability through a trustworthy data-adaptive method instead of a heuristic way.
- Regime-switching differentiable particle filters
In real-world applications, both dynamic and measurement models in state-spaces models (SSMs) can be constituted by a variety of modes and be switchable. For instance, in object tracking, targets can exhibit diverse status, e.g., people may stroll gently, ride in heavy traffic, or cruise on motorways at different times. On the other hand, the observational measurement can be susceptible to various conditions such as obstacle sheltering or weather evolving. Therefore, it leads to an additional complexity layer of the regime mixture uncertainty in SSMs, i.e., regime-switching state-space models. This project leads to analyse such complicated dynamical systems with satisfied flexibility and scalability through a trustworthy data-adaptive method instead of a heuristic way.
University roles and responsibilities
- Computer Science PGR Representative (2022/23)
Teaching
- Demonstrator for COM2028 Artificial Intelligence (2023/02-2023/06)
Publications
Differentiable particle filters are an emerging class of particle filtering methods that use neural networks to construct and learn parametric state-space models. In real-world applications, both the state dynamics and measurements can switch between a set of candidate models. For instance, in target tracking, vehicles can idle, move through traffic, or cruise on motorways, and measurements are collected in different geographical or weather conditions. This paper proposes a new differentiable particle filter for regime-switching state-space models. The method can learn a set of unknown candidate dynamic and measurement models and track the state posteriors. We evaluate the performance of the novel algorithm in relevant models, showing its great performance compared to other competitive algorithms.