Robust cognitive load detection from wrist-band sensors

In recent years, the detection of cognitive load has received a lot of attention. Understanding the circumstances in which cognitive load occurs and reliably predicting such occurrences, offers the potential for considerable advances in the field of Human-Computer Interaction (HCI). Numerous HCI app...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Vadim Borisov, Enkelejda Kasneci, Gjergji Kasneci
Formato: article
Lenguaje:EN
Publicado: Elsevier 2021
Materias:
Acceso en línea:https://doaj.org/article/def24efabc244de1895efba7ce759274
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:def24efabc244de1895efba7ce759274
record_format dspace
spelling oai:doaj.org-article:def24efabc244de1895efba7ce7592742021-12-01T05:04:23ZRobust cognitive load detection from wrist-band sensors2451-958810.1016/j.chbr.2021.100116https://doaj.org/article/def24efabc244de1895efba7ce7592742021-08-01T00:00:00Zhttp://www.sciencedirect.com/science/article/pii/S2451958821000646https://doaj.org/toc/2451-9588In recent years, the detection of cognitive load has received a lot of attention. Understanding the circumstances in which cognitive load occurs and reliably predicting such occurrences, offers the potential for considerable advances in the field of Human-Computer Interaction (HCI). Numerous HCI applications, ranging from medical and health-related solutions to (smart) automotive environments, would directly benefit from the reliable detection of cognitive load. However, this task still remains highly challenging. We present a machine learning (ML) approach based on ensemble learning for robust cognitive load classification. The features used by the proposed solution are generated from the interpretation of physiological measurements (e.g., heart rate, r-r interval, skin temperature, and skin response) from a wearable device. Hence, our approach consists of two steps: (1) transforming the original data into discriminative features and (2) training an ensemble model to accurately and robustly predict cognitive load. The empirical results confirm that our method has a superior performance compared to various state-of-the-art baselines on the original and transformed data. Moreover, in the open-data CogLoad@UbiComp 2020 Competition, the proposed approach achieved the best results among 17 competing approaches and outperformed all participating competitors by a considerable margin.Vadim BorisovEnkelejda KasneciGjergji KasneciElsevierarticleCognitive load detectionHuman-computer interactionMachine learningEnsemble methodsElectronic computers. Computer scienceQA75.5-76.95PsychologyBF1-990ENComputers in Human Behavior Reports, Vol 4, Iss , Pp 100116- (2021)
institution DOAJ
collection DOAJ
language EN
topic Cognitive load detection
Human-computer interaction
Machine learning
Ensemble methods
Electronic computers. Computer science
QA75.5-76.95
Psychology
BF1-990
spellingShingle Cognitive load detection
Human-computer interaction
Machine learning
Ensemble methods
Electronic computers. Computer science
QA75.5-76.95
Psychology
BF1-990
Vadim Borisov
Enkelejda Kasneci
Gjergji Kasneci
Robust cognitive load detection from wrist-band sensors
description In recent years, the detection of cognitive load has received a lot of attention. Understanding the circumstances in which cognitive load occurs and reliably predicting such occurrences, offers the potential for considerable advances in the field of Human-Computer Interaction (HCI). Numerous HCI applications, ranging from medical and health-related solutions to (smart) automotive environments, would directly benefit from the reliable detection of cognitive load. However, this task still remains highly challenging. We present a machine learning (ML) approach based on ensemble learning for robust cognitive load classification. The features used by the proposed solution are generated from the interpretation of physiological measurements (e.g., heart rate, r-r interval, skin temperature, and skin response) from a wearable device. Hence, our approach consists of two steps: (1) transforming the original data into discriminative features and (2) training an ensemble model to accurately and robustly predict cognitive load. The empirical results confirm that our method has a superior performance compared to various state-of-the-art baselines on the original and transformed data. Moreover, in the open-data CogLoad@UbiComp 2020 Competition, the proposed approach achieved the best results among 17 competing approaches and outperformed all participating competitors by a considerable margin.
format article
author Vadim Borisov
Enkelejda Kasneci
Gjergji Kasneci
author_facet Vadim Borisov
Enkelejda Kasneci
Gjergji Kasneci
author_sort Vadim Borisov
title Robust cognitive load detection from wrist-band sensors
title_short Robust cognitive load detection from wrist-band sensors
title_full Robust cognitive load detection from wrist-band sensors
title_fullStr Robust cognitive load detection from wrist-band sensors
title_full_unstemmed Robust cognitive load detection from wrist-band sensors
title_sort robust cognitive load detection from wrist-band sensors
publisher Elsevier
publishDate 2021
url https://doaj.org/article/def24efabc244de1895efba7ce759274
work_keys_str_mv AT vadimborisov robustcognitiveloaddetectionfromwristbandsensors
AT enkelejdakasneci robustcognitiveloaddetectionfromwristbandsensors
AT gjergjikasneci robustcognitiveloaddetectionfromwristbandsensors
_version_ 1718405573016289280