Estimating Change Intensity and Duration in Human Activity Recognition using Martingales

Jonathan Etumusei; Jorge Martinez Carracedo; Sally I McClean

Estimating Change Intensity and Duration in Human Activity Recognition using Martingales

Jonathan Etumusei, Jorge Martinez Carracedo, Sally I McClean

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The subject of physical activity is becoming prominent in the healthcare system for the improvement and monitoring of movement disabilities. Existing algorithms are effective in detecting changes in data streams but most of these approaches are not focused on measuring the change intensity and duration. In this paper, we improve on the geometric moving average martingale method by optimising the parameters in the weighted average using a genetic algorithm. The proposed approach enables us to estimate the intensity and duration of transitions that happen in human activity recognition scenarios. Results show that the proposed method makes some improvement over previous martingale techniques.

Original language	English
Title of host publication	The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science
Subtitle of host publication	LOD 2021: Machine Learning, Optimization, and Data Science
Publisher	Springer
Pages	553-567
Number of pages	14
Volume	LNCS, volume 13163
ISBN (Electronic)	978-3-030-95467-3
ISBN (Print)	978-3-030-95466-6
Publication status	Published - 2 Feb 2022
Event	The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science - Grasmere England, Lake Districk, United Kingdom Duration: 4 Oct 2021 → 8 Oct 2021 https://lod2021.icas.cc/

Conference

Conference	The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science
Abbreviated title	LOD2021
Country/Territory	United Kingdom
City	Lake Districk
Period	4/10/21 → 8/10/21
Internet address	https://lod2021.icas.cc/

Keywords

Change detection
martingales
human activity recognition

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

https://link.springer.com/chapter/10.1007/978-3-030-95467-3_40

Cite this

Etumusei, J., Martinez Carracedo, J., & McClean, S. I. (2022). Estimating Change Intensity and Duration in Human Activity Recognition using Martingales. In The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science : LOD 2021: Machine Learning, Optimization, and Data Science (Vol. LNCS, volume 13163, pp. 553-567). Springer. https://link.springer.com/chapter/10.1007/978-3-030-95467-3_40

Etumusei, Jonathan ; Martinez Carracedo, Jorge ; McClean, Sally I. / Estimating Change Intensity and Duration in Human Activity Recognition using Martingales. The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science : LOD 2021: Machine Learning, Optimization, and Data Science. Vol. LNCS, volume 13163 Springer, 2022. pp. 553-567

@inproceedings{99e15187ac7042c1bbde474203fdb1e9,

title = "Estimating Change Intensity and Duration in Human Activity Recognition using Martingales",

abstract = "The subject of physical activity is becoming prominent in the healthcare system for the improvement and monitoring of movement disabilities. Existing algorithms are effective in detecting changes in data streams but most of these approaches are not focused on measuring the change intensity and duration. In this paper, we improve on the geometric moving average martingale method by optimising the parameters in the weighted average using a genetic algorithm. The proposed approach enables us to estimate the intensity and duration of transitions that happen in human activity recognition scenarios. Results show that the proposed method makes some improvement over previous martingale techniques.",

keywords = "Change detection, martingales, human activity recognition",

author = "Jonathan Etumusei and {Martinez Carracedo}, Jorge and McClean, {Sally I}",

year = "2022",

month = feb,

day = "2",

language = "English",

isbn = "978-3-030-95466-6",

volume = "LNCS, volume 13163",

pages = "553--567",

booktitle = "The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science",

publisher = "Springer",

address = "Switzerland",

note = "The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science , LOD2021 ; Conference date: 04-10-2021 Through 08-10-2021",

url = "https://lod2021.icas.cc/",

}

Etumusei, J , Martinez Carracedo, J & McClean, SI 2022, Estimating Change Intensity and Duration in Human Activity Recognition using Martingales. in The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science : LOD 2021: Machine Learning, Optimization, and Data Science. vol. LNCS, volume 13163, Springer, pp. 553-567, The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science , Lake Districk, United Kingdom, 4/10/21. <https://link.springer.com/chapter/10.1007/978-3-030-95467-3_40>

Estimating Change Intensity and Duration in Human Activity Recognition using Martingales. / Etumusei, Jonathan ; Martinez Carracedo, Jorge ; McClean, Sally I.

The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science : LOD 2021: Machine Learning, Optimization, and Data Science. Vol. LNCS, volume 13163 Springer, 2022. p. 553-567.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Estimating Change Intensity and Duration in Human Activity Recognition using Martingales

AU - Etumusei, Jonathan

AU - Martinez Carracedo, Jorge

AU - McClean, Sally I

PY - 2022/2/2

Y1 - 2022/2/2

N2 - The subject of physical activity is becoming prominent in the healthcare system for the improvement and monitoring of movement disabilities. Existing algorithms are effective in detecting changes in data streams but most of these approaches are not focused on measuring the change intensity and duration. In this paper, we improve on the geometric moving average martingale method by optimising the parameters in the weighted average using a genetic algorithm. The proposed approach enables us to estimate the intensity and duration of transitions that happen in human activity recognition scenarios. Results show that the proposed method makes some improvement over previous martingale techniques.

AB - The subject of physical activity is becoming prominent in the healthcare system for the improvement and monitoring of movement disabilities. Existing algorithms are effective in detecting changes in data streams but most of these approaches are not focused on measuring the change intensity and duration. In this paper, we improve on the geometric moving average martingale method by optimising the parameters in the weighted average using a genetic algorithm. The proposed approach enables us to estimate the intensity and duration of transitions that happen in human activity recognition scenarios. Results show that the proposed method makes some improvement over previous martingale techniques.

KW - Change detection

KW - martingales

KW - human activity recognition

M3 - Conference contribution

SN - 978-3-030-95466-6

VL - LNCS, volume 13163

SP - 553

EP - 567

BT - The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science

PB - Springer

T2 - The 7th International Online & Onsite Conference on Machine Learning, Optimization, and Data Science

Y2 - 4 October 2021 through 8 October 2021

ER -

Estimating Change Intensity and Duration in Human Activity Recognition using Martingales

Abstract

Conference

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this