The INI has a new website!

This is a legacy webpage. Please visit the new site to ensure you are seeing up to date information.

Isaac Newton Institute for Mathematical Sciences

Efficient Analysis of Changepoint Models using Dynamic Programming Methods.

Presenter: Robert Maidstone (STOR-i CDT, Lancaster University)

Co-authors: Prof. Paul Fearnhead (STOR-i CDT, Lancaster University), Prof. Adam Letchford (STOR-i CDT, Lancaster University)

Abstract

When detecting multiple changepoints in large datasets optimally, often the computational time taken by the detection method becomes an issue. While heuristic methods can be as fast as $\mathcal{O}(n \log n)$, exact methods tend to be much slower. This poster will concentrated on Dynamic Programming based algorithms, which tend to be $\mathcal{O}(n^2)$, and discuss how these can sped up by pruning (such as in Killick et al. (2011) and Rigaill (2010), as well as using the authors' own methods) to compete with heuristic approaches.

Related Links