Real-time clustering of noisy time series

Lead Research Organisation: University of Manchester
Department Name: Mathematics

Abstract

The PhD project is concerned with the development of new approaches for clustering of noisy time series, with an application to real-time measurements being taken in large industrial assets. Firstly, the student will investigate low-dimensional representations of time series which preserve relevant features and allow for automatic classification using techniques from machine learning, such as support vector machines. Secondly, the real-time aspect of data collection will necessitate the need for efficient updating techniques in the linear algebra routines underlying the training and assignment phases.

Publications

10 25 50

publication icon
Steven Elsworth (2020) The block rational Arnoldi method

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/N509565/1 01/10/2016 30/09/2021
1775511 Studentship EP/N509565/1 01/09/2016 31/03/2020 Steven Elsworth
 
Description - Introduced a method to convert between different representations of rational interpolant.
- Developed the theory behind block rational Krylov spaces.
- Established a connection between vector autoregressive time series models and block polynomial Arnoldi decompositions.
- Introduced a new symbolic representation of time series.
Exploitation Route Block rational Krylov spaces have many applications including solving eigenvalue problems, matrix equations and model order reduction to name just a few. The new symbolic representation, which we call ABBA, has a variety of applications in the time series data mining community.
Sectors Other