Ensemble classifier of long short-term memory with fuzzy temporal windows on binary sensors for activity recognition

Javier Medina-Quero, Shuai Zhang, CD Nugent, Macarena Espinilla

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

There are approaches that successfully recognize activities of daily living by using a trained classifier on feature vectors created from binary sensor data. Although these approaches have been successful, there are still open issues such as the evaluation of multiple temporal windows, ensembles of classifiers or unbalanced classes which need to be addressed in order to improve the performance of the real-time activity recognition process. In this paper, we present a methodology for Real-Time Activity Recognition based on the diverse fields of Machine Learning, including Fuzzy Logic and Recurrent Neural Networks. The methodology uses a long-term and short-term representation of binary-sensor activations based on Fuzzy Temporal Windows. The paper proposes an ensemble of activity-based classifiers for the purposes of balanced training, where each classifier in the ensemble is a Long Short-Term Memory. The approach was evaluated using two binary-sensor datasets of daily living activities and benchmarked against previous approaches based on the combination of sensor activation features.
LanguageEnglish
Pages441-453
Number of pages13
JournalExpert Systems with Applications
Volume114
Early online date1 Aug 2018
DOIs
Publication statusE-pub ahead of print - 1 Aug 2018

Fingerprint

Classifiers
Sensors
Chemical activation
Recurrent neural networks
Fuzzy logic
Learning systems
Long short-term memory

Keywords

  • Activity recognition
  • Fuzzy temporal windows
  • Long short-term memory
  • Unbalanced data
  • Ensemble architectures

Cite this

@article{53bbcf2ebec2410b8c66f2de5bbc5852,
title = "Ensemble classifier of long short-term memory with fuzzy temporal windows on binary sensors for activity recognition",
abstract = "There are approaches that successfully recognize activities of daily living by using a trained classifier on feature vectors created from binary sensor data. Although these approaches have been successful, there are still open issues such as the evaluation of multiple temporal windows, ensembles of classifiers or unbalanced classes which need to be addressed in order to improve the performance of the real-time activity recognition process. In this paper, we present a methodology for Real-Time Activity Recognition based on the diverse fields of Machine Learning, including Fuzzy Logic and Recurrent Neural Networks. The methodology uses a long-term and short-term representation of binary-sensor activations based on Fuzzy Temporal Windows. The paper proposes an ensemble of activity-based classifiers for the purposes of balanced training, where each classifier in the ensemble is a Long Short-Term Memory. The approach was evaluated using two binary-sensor datasets of daily living activities and benchmarked against previous approaches based on the combination of sensor activation features.",
keywords = "Activity recognition, Fuzzy temporal windows, Long short-term memory, Unbalanced data, Ensemble architectures",
author = "Javier Medina-Quero and Shuai Zhang and CD Nugent and Macarena Espinilla",
year = "2018",
month = "8",
day = "1",
doi = "10.1016/j.eswa.2018.07.068",
language = "English",
volume = "114",
pages = "441--453",
journal = "Expert Systems with Applications",
issn = "0957-4174",
publisher = "Elsevier",

}

Ensemble classifier of long short-term memory with fuzzy temporal windows on binary sensors for activity recognition. / Medina-Quero, Javier; Zhang, Shuai; Nugent, CD; Espinilla, Macarena.

In: Expert Systems with Applications, Vol. 114, 01.08.2018, p. 441-453.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Ensemble classifier of long short-term memory with fuzzy temporal windows on binary sensors for activity recognition

AU - Medina-Quero, Javier

AU - Zhang, Shuai

AU - Nugent, CD

AU - Espinilla, Macarena

PY - 2018/8/1

Y1 - 2018/8/1

N2 - There are approaches that successfully recognize activities of daily living by using a trained classifier on feature vectors created from binary sensor data. Although these approaches have been successful, there are still open issues such as the evaluation of multiple temporal windows, ensembles of classifiers or unbalanced classes which need to be addressed in order to improve the performance of the real-time activity recognition process. In this paper, we present a methodology for Real-Time Activity Recognition based on the diverse fields of Machine Learning, including Fuzzy Logic and Recurrent Neural Networks. The methodology uses a long-term and short-term representation of binary-sensor activations based on Fuzzy Temporal Windows. The paper proposes an ensemble of activity-based classifiers for the purposes of balanced training, where each classifier in the ensemble is a Long Short-Term Memory. The approach was evaluated using two binary-sensor datasets of daily living activities and benchmarked against previous approaches based on the combination of sensor activation features.

AB - There are approaches that successfully recognize activities of daily living by using a trained classifier on feature vectors created from binary sensor data. Although these approaches have been successful, there are still open issues such as the evaluation of multiple temporal windows, ensembles of classifiers or unbalanced classes which need to be addressed in order to improve the performance of the real-time activity recognition process. In this paper, we present a methodology for Real-Time Activity Recognition based on the diverse fields of Machine Learning, including Fuzzy Logic and Recurrent Neural Networks. The methodology uses a long-term and short-term representation of binary-sensor activations based on Fuzzy Temporal Windows. The paper proposes an ensemble of activity-based classifiers for the purposes of balanced training, where each classifier in the ensemble is a Long Short-Term Memory. The approach was evaluated using two binary-sensor datasets of daily living activities and benchmarked against previous approaches based on the combination of sensor activation features.

KW - Activity recognition

KW - Fuzzy temporal windows

KW - Long short-term memory

KW - Unbalanced data

KW - Ensemble architectures

U2 - 10.1016/j.eswa.2018.07.068

DO - 10.1016/j.eswa.2018.07.068

M3 - Article

VL - 114

SP - 441

EP - 453

JO - Expert Systems with Applications

T2 - Expert Systems with Applications

JF - Expert Systems with Applications

SN - 0957-4174

ER -