[CfP] EUSIPCO 2022 Special Session on Machine Learning for Sequential Monte Carlo Methods

Skip to first unread message


Jan 24, 2022, 9:23:30 AM1/24/22
to Machine Learning News
We are organizing a Special Session on “Machine Learning for Sequential Monte Carlo Methods” at the 30th European Signal Processing Conference (EUSIPCO 2022) in Belgrade, Serbia, on 29th August – 2nd September 2022. We hope to bring together the best research on this topic (more details below). Please consider submitting your latest work to our session (see submission instructions on the EUSIPCO 2022 website https://2022.eusipco.org/).
Important dates
20th February 2022: Full paper submission
6th May 2022: Notification of acceptance
5th June 2022: Camera-Ready Paper Submission

Yunpeng Li (University of Surrey, UK)
Simon Maskell (University of Liverpool, UK)
Uwe D. Hanebeck (Karlsruhe Institute of Technology (KIT), Germany)
Tiancheng Li (Northwestern Polytechnical University, China)
Rico Jonschkowski (Google, USA)

Sequential Monte Carlo (SMC) methods, such as particle filters and SMC samplers, are a family of flexible simulation-based algorithms to compute arbitrary probability distributions involved in sequential and non-sequential estimation and sampling tasks. The methods are statistically consistent and widely used in applications including computer vision, target tracking, finance, navigation, and robotics. A particular challenge for the deployment of SMC methods is the need to specify the often nonlinear models that simulate state dynamics and their relation to measurements. This becomes non-trivial for practitioners when dealing with complex environments and big data.

This Special Session focuses on recent advances in developing machine learning approaches to automatically build various components of SMC. Replacing heuristic models in SMC with data-driven ones would make them an extremely powerful tool in real-world applications. The Special Session also welcomes contributions on closely related areas, including the interplay between machine learning and SMC as well as comparisons with other nonlinear filtering methods. It aims to bring together researchers working on the intersection of dynamic systems, signal processing and machine learning, facilitating discussions of research questions and identifying promising future directions.

Topics of Interests
Topics of this Special Session may include, but are not limited to the following aspects:
Modeling and prediction of object dynamics.
Learning of likelihood and measurement models.
Automatically adapting proposal distributions.
Supervised, semi-supervised, and unsupervised loss functions.
Differentiable resampling methods.
Online learning and continual learning.
Bridging between SMC and deep learning methods for temporal sequences.
Physics-inspired methods for SMC methods (transport methods, simulated annealing, particle flow etc.).
Machine learning with SMC samplers.
Comparison between SMC and other machine learning-based nonlinear filtering approaches (RNNs, deterministic particle filters, etc.)
Gaussian Process for nonlinear filter identification
Performance evaluation of estimation methods.
Applications of SMC methods.

Reply all
Reply to author
0 new messages