An Optimal Linear Time Algorithm for Quasi-Monotonic Segmentation
Abstract
Monotonicity is a simple yet significant qualitative characteristic. We consider the problem of segmenting an array in up to K segments. We want segments to be as monotonic as possible and to alternate signs. We propose a quality metric for this problem, present an optimal linear time algorithm based on novel formalism, and compare experimentally its performance to a linear time top-down regression algorithm. We show that our algorithm is faster and more accurate. Applications include pattern recognition and qualitative modeling.
Keywords
Piecewise Quasi-Monotone Functions, Segmentation, ECG, Spline
Reference
Daniel Lemire, Martin Brooks, Yuhong Yan, An Optimal Linear Time Algorithm for Quasi-Monotonic Segmentation, In Proceedings of IEEE Data Mining 2005 (ICDM-05), November 2005.
Download
Hint : It is sometimes necessary to hold down shift while clicking in order to save a document.
Software
The software used for this paper is available from Daniel Lemire upon request.
BibTeX
@inproceedings{YLBICDMO05,
author = {Daniel Lemire and Martin Brooks and Yuhong Yan},
title = {An Optimal Linear Time Algorithm for Quasi-Monotonic Segmentation},
booktitle = {ICDM'05},
pages={709--712},
year = {2005}
}
Authors
- Daniel Lemire: lemire at acm.org
- Yuhong Yan: yuhong.yan@nrc.gc.ca
- Martin Brooks: martin.brooks@nrc.gc.ca
Related work
- Martin Brooks, Yuhong Yan, Daniel Lemire, Scale-Based Monotonicity Analysis in Qualitative Modelling with Flat Segments, IJCAI05, Edinburgh, UK, July 2005.
- Daniel Lemire's publications