Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration

Jaakko Tervonen; Rajdeep Kumar Nath; Kati Pettersson; Johanna Närväinen; Jani Mäntyjärvi

doi:10.1145/3594739.3610731

Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration

Jaakko Tervonen, Rajdeep Kumar Nath, Kati Pettersson, Johanna Närväinen, Jani Mäntyjärvi

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

2 Citations (Scopus)

Abstract

Human physiology and reactions to external stimuli differ between individuals. Researchers have developed strategies to adapt to these differences but the adaptations generally require data from each individual beforehand. A cold-start occurs when there is no data from a new individual. To address this, current study proposes user calibration which uses short segments of easily obtainable baseline data to adapt to new individuals. Experiments were conducted on two public stress and affect detection datasets, WESAD and SWELL-KW, to assess the effectiveness of the proposed calibration method and to determine suitable duration of the baseline measurement. Results showed that user calibration always beat the non-personalized model and segments of 3-8 minutes seemed to be most promising to consider for future use.

Original language	English
Title of host publication	Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23
Publisher	Association for Computing Machinery ACM
Pages	417-422
Number of pages	6
ISBN (Electronic)	979-8-4007-0200-6
DOIs	https://doi.org/10.1145/3594739.3610731
Publication status	Published - 8 Oct 2023
MoE publication type	A4 Article in a conference publication
Event	2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23 - Cancun, Mexico Duration: 8 Oct 2023 → 12 Oct 2023

Conference

Conference	2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23
Country/Territory	Mexico
City	Cancun
Period	8/10/23 → 12/10/23

Funding

The work was funded by the Academy of Finland under GrantNos.: 334092, 351282 and VTT.

Keywords

personalization
stress detection
wearable devices
wellbeing

Access to Document

10.1145/3594739.3610731Licence: CC BY

Cite this

Tervonen, J., Nath, R. K., Pettersson, K., Närväinen, J., & Mäntyjärvi, J. (2023). Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration. In Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23 (pp. 417-422). Association for Computing Machinery ACM. https://doi.org/10.1145/3594739.3610731

Tervonen, Jaakko ; Nath, Rajdeep Kumar ; Pettersson, Kati et al. / Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration. Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23. Association for Computing Machinery ACM, 2023. pp. 417-422

@inproceedings{70106a7518d44077943bfb8511e17c2d,

title = "Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration",

abstract = "Human physiology and reactions to external stimuli differ between individuals. Researchers have developed strategies to adapt to these differences but the adaptations generally require data from each individual beforehand. A cold-start occurs when there is no data from a new individual. To address this, current study proposes user calibration which uses short segments of easily obtainable baseline data to adapt to new individuals. Experiments were conducted on two public stress and affect detection datasets, WESAD and SWELL-KW, to assess the effectiveness of the proposed calibration method and to determine suitable duration of the baseline measurement. Results showed that user calibration always beat the non-personalized model and segments of 3-8 minutes seemed to be most promising to consider for future use.",

keywords = "personalization, stress detection, wearable devices, wellbeing",

author = "Jaakko Tervonen and Nath, {Rajdeep Kumar} and Kati Pettersson and Johanna N{\"a}rv{\"a}inen and Jani M{\"a}ntyj{\"a}rvi",

year = "2023",

month = oct,

day = "8",

doi = "10.1145/3594739.3610731",

language = "English",

pages = "417--422",

booktitle = "Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC {\textquoteright}23",

publisher = "Association for Computing Machinery ACM",

address = "United States",

note = "2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC {\textquoteright}23 ; Conference date: 08-10-2023 Through 12-10-2023",

}

Tervonen, J , Nath, RK , Pettersson, K , Närväinen, J & Mäntyjärvi, J 2023, Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration. in Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23. Association for Computing Machinery ACM, pp. 417-422, 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23, Cancun, Mexico, 8/10/23. https://doi.org/10.1145/3594739.3610731

Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration. / Tervonen, Jaakko ; Nath, Rajdeep Kumar ; Pettersson, Kati et al.
Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23. Association for Computing Machinery ACM, 2023. p. 417-422.

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration

AU - Tervonen, Jaakko

AU - Nath, Rajdeep Kumar

AU - Pettersson, Kati

AU - Närväinen, Johanna

AU - Mäntyjärvi, Jani

PY - 2023/10/8

Y1 - 2023/10/8

N2 - Human physiology and reactions to external stimuli differ between individuals. Researchers have developed strategies to adapt to these differences but the adaptations generally require data from each individual beforehand. A cold-start occurs when there is no data from a new individual. To address this, current study proposes user calibration which uses short segments of easily obtainable baseline data to adapt to new individuals. Experiments were conducted on two public stress and affect detection datasets, WESAD and SWELL-KW, to assess the effectiveness of the proposed calibration method and to determine suitable duration of the baseline measurement. Results showed that user calibration always beat the non-personalized model and segments of 3-8 minutes seemed to be most promising to consider for future use.

AB - Human physiology and reactions to external stimuli differ between individuals. Researchers have developed strategies to adapt to these differences but the adaptations generally require data from each individual beforehand. A cold-start occurs when there is no data from a new individual. To address this, current study proposes user calibration which uses short segments of easily obtainable baseline data to adapt to new individuals. Experiments were conducted on two public stress and affect detection datasets, WESAD and SWELL-KW, to assess the effectiveness of the proposed calibration method and to determine suitable duration of the baseline measurement. Results showed that user calibration always beat the non-personalized model and segments of 3-8 minutes seemed to be most promising to consider for future use.

KW - personalization

KW - stress detection

KW - wearable devices

KW - wellbeing

UR - http://www.scopus.com/inward/record.url?scp=85175460285&partnerID=8YFLogxK

U2 - 10.1145/3594739.3610731

DO - 10.1145/3594739.3610731

M3 - Conference article in proceedings

SP - 417

EP - 422

BT - Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23

PB - Association for Computing Machinery ACM

T2 - 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23

Y2 - 8 October 2023 through 12 October 2023

ER -

Tervonen J , Nath RK , Pettersson K , Närväinen J , Mäntyjärvi J. Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration. In Adjunct Proceedings of the 2023 ACM International Joint Conference on Pervasive and Ubiquitous Computing & the 2023 ACM International Symposium on Wearable Computing, UbiComp/ISWC ’23. Association for Computing Machinery ACM. 2023. p. 417-422 doi: 10.1145/3594739.3610731

Cold-Start Model Adaptation: Evaluation of Short Baseline Calibration

Abstract

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this