Comparing Person-Specific and Independent Models on Subject-Dependent and Independent Human Activity Recognition Performance
The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess whether HAR models really do achieve better subject-dependent performance than subject-independent performance, whether a model trained with data...
Gespeichert in:
| Veröffentlicht in: | Sensors (Basel, Switzerland) Jg. 20; H. 13; S. 3647 |
|---|---|
| Hauptverfasser: | , , , |
| Format: | Journal Article |
| Sprache: | Englisch |
| Veröffentlicht: |
Switzerland
MDPI
29.06.2020
MDPI AG |
| Schlagworte: | |
| ISSN: | 1424-8220, 1424-8220 |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Abstract | The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess whether HAR models really do achieve better subject-dependent performance than subject-independent performance, whether a model trained with data from many users achieves better subject-independent performance than one trained with data from a single person, and whether one trained with data from a single specific target user performs better for that user than one trained with data from many. To those ends, we compare four popular machine learning algorithms’ subject-dependent and subject-independent performances across eight datasets using three different personalisation–generalisation approaches, which we term person-independent models (PIMs), person-specific models (PSMs), and ensembles of PSMs (EPSMs). We further consider three different ways to construct such an ensemble: unweighted, κ -weighted, and baseline-feature-weighted. Our analysis shows that PSMs outperform PIMs by 43.5% in terms of their subject-dependent performances, whereas PIMs outperform PSMs by 55.9% and κ -weighted EPSMs—the best-performing EPSM type—by 16.4% in terms of the subject-independent performance. |
|---|---|
| AbstractList | The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess whether HAR models really do achieve better subject-dependent performance than subject-independent performance, whether a model trained with data from many users achieves better subject-independent performance than one trained with data from a single person, and whether one trained with data from a single specific target user performs better for that user than one trained with data from many. To those ends, we compare four popular machine learning algorithms’ subject-dependent and subject-independent performances across eight datasets using three different personalisation–generalisation approaches, which we term person-independent models (PIMs), person-specific models (PSMs), and ensembles of PSMs (EPSMs). We further consider three different ways to construct such an ensemble: unweighted, κ -weighted, and baseline-feature-weighted. Our analysis shows that PSMs outperform PIMs by 43.5% in terms of their subject-dependent performances, whereas PIMs outperform PSMs by 55.9% and κ -weighted EPSMs—the best-performing EPSM type—by 16.4% in terms of the subject-independent performance. The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess whether HAR models really do achieve better subject-dependent performance than subject-independent performance, whether a model trained with data from many users achieves better subject-independent performance than one trained with data from a single person, and whether one trained with data from a single specific target user performs better for that user than one trained with data from many. To those ends, we compare four popular machine learning algorithms' subject-dependent and subject-independent performances across eight datasets using three different personalisation-generalisation approaches, which we term person-independent models (PIMs), person-specific models (PSMs), and ensembles of PSMs (EPSMs). We further consider three different ways to construct such an ensemble: unweighted, κ -weighted, and baseline-feature-weighted. Our analysis shows that PSMs outperform PIMs by 43.5% in terms of their subject-dependent performances, whereas PIMs outperform PSMs by 55.9% and κ -weighted EPSMs-the best-performing EPSM type-by 16.4% in terms of the subject-independent performance.The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess whether HAR models really do achieve better subject-dependent performance than subject-independent performance, whether a model trained with data from many users achieves better subject-independent performance than one trained with data from a single person, and whether one trained with data from a single specific target user performs better for that user than one trained with data from many. To those ends, we compare four popular machine learning algorithms' subject-dependent and subject-independent performances across eight datasets using three different personalisation-generalisation approaches, which we term person-independent models (PIMs), person-specific models (PSMs), and ensembles of PSMs (EPSMs). We further consider three different ways to construct such an ensemble: unweighted, κ -weighted, and baseline-feature-weighted. Our analysis shows that PSMs outperform PIMs by 43.5% in terms of their subject-dependent performances, whereas PIMs outperform PSMs by 55.9% and κ -weighted EPSMs-the best-performing EPSM type-by 16.4% in terms of the subject-independent performance. |
| Author | O’Flynn, Brendan Tedesco, Salvatore Brown, Kenneth N. Scheurer, Sebastian |
| AuthorAffiliation | 1 Insight Centre for Data Analytics, School of Computer Science and Information Technology, University College Cork, T12 XF62 Cork, Ireland; brendan.oflynn@tyndall.ie 3 CONNECT Centre for Future Networks and Communications, Tyndall National Institute, University College Cork, T12 R5CP Cork, Ireland 2 Tyndall National Institute, University College Cork, T12 R5CP Cork, Ireland; salvatore.tedesco@tyndall.ie |
| AuthorAffiliation_xml | – name: 3 CONNECT Centre for Future Networks and Communications, Tyndall National Institute, University College Cork, T12 R5CP Cork, Ireland – name: 1 Insight Centre for Data Analytics, School of Computer Science and Information Technology, University College Cork, T12 XF62 Cork, Ireland; brendan.oflynn@tyndall.ie – name: 2 Tyndall National Institute, University College Cork, T12 R5CP Cork, Ireland; salvatore.tedesco@tyndall.ie |
| Author_xml | – sequence: 1 givenname: Sebastian orcidid: 0000-0002-2163-2168 surname: Scheurer fullname: Scheurer, Sebastian – sequence: 2 givenname: Salvatore orcidid: 0000-0002-7752-2240 surname: Tedesco fullname: Tedesco, Salvatore – sequence: 3 givenname: Brendan orcidid: 0000-0002-5522-2597 surname: O’Flynn fullname: O’Flynn, Brendan – sequence: 4 givenname: Kenneth N. surname: Brown fullname: Brown, Kenneth N. |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/32610614$$D View this record in MEDLINE/PubMed |
| BookMark | eNplkktv1DAUhS1URB-w4A-gLGER6tck8QapmtJ2pCIQhbXlXN8MHiV2sJNKlfjxeJh21JaNbfmc-x35-h6TAx88EvKW0Y9CKHqaOGVCVLJ-QY6Y5LJsOKcHj86H5DilDaVcCNG8IoeCV4xWTB6RP8swjCY6vy6-YUzBlzcjguscFMbbYuUtjpgXPxVfgsU-FcEXN3O7QZjK87323Hs1D8YXZzC5WzfdFd8Rwtq7yeXiHNOFmGXA1-RlZ_qEb-73E_Lz4vOP5VV5_fVytTy7LkFKNpWA2HEqKt4aoWzdsQVysFDDwqJquepsjaCAYStaidC0tQIjJeVSWUbbVpyQ1Y5rg9noMbrBxDsdjNP_LkJcaxMnBz1qMKKzVddaECp3r1ZsYRvT2E5UAI3csj7tWOPcDmghvzaa_gn0qeLdL70Ot7oWtRSsyoD394AYfs-YJj24BNj3xmOYk-aSqZoqJXi2vnuctQ95-L5sON0ZIIaUInYa3GS2bc7RrteM6u2A6P2A5IoPzyoeoP97_wJqtb5O |
| CitedBy_id | crossref_primary_10_3390_app11167660 crossref_primary_10_3390_app13031512 crossref_primary_10_3390_s24113697 crossref_primary_10_3390_s22083094 |
| Cites_doi | 10.1007/978-3-540-24646-6_1 10.25080/Majora-92bf1922-00a 10.1145/2499621 10.1145/3361684.3361689 10.1109/PERCOM.2016.7456521 10.1109/HealthCom.2018.8531148 10.1109/ICMLA.2015.48 10.1109/SMC.2015.263 10.1109/TITB.2005.856864 10.1016/j.patrec.2012.12.014 10.1109/MCSE.2011.37 10.3390/s140610146 10.18637/jss.v067.i01 10.1109/IJCNN.2019.8852299 10.1017/CBO9780511790942 10.3390/s16010115 10.1109/ISWC.2012.13 10.1016/j.asoc.2015.01.025 10.1109/HealthCom.2016.7749439 10.1109/TPAMI.2015.2389824 10.1016/j.patrec.2018.02.010 10.1109/BSN.2017.7935994 10.3390/s140610691 |
| ContentType | Journal Article |
| Copyright | 2020 by the authors. 2020 |
| Copyright_xml | – notice: 2020 by the authors. 2020 |
| DBID | AAYXX CITATION CGR CUY CVF ECM EIF NPM 7X8 5PM DOA |
| DOI | 10.3390/s20133647 |
| DatabaseName | CrossRef Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed MEDLINE - Academic PubMed Central (Full Participant titles) DOAJ Directory of Open Access Journals |
| DatabaseTitle | CrossRef MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) MEDLINE - Academic |
| DatabaseTitleList | CrossRef MEDLINE - Academic MEDLINE |
| Database_xml | – sequence: 1 dbid: DOA name: Directory of Open Access Journals (DOAJ) url: https://www.doaj.org/ sourceTypes: Open Website – sequence: 2 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 3 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Engineering |
| EISSN | 1424-8220 |
| ExternalDocumentID | oai_doaj_org_article_ca3fd6fbdc394247915d8a8df36cc84b PMC7374316 32610614 10_3390_s20133647 |
| Genre | Journal Article |
| GrantInformation_xml | – fundername: Enterprise Ireland grantid: IR20140024 – fundername: Science Foundation Ireland grantid: 12/RC/2289-P2 – fundername: European Regional Development Fund grantid: 13/RC/2077-CONNECT |
| GroupedDBID | --- 123 2WC 53G 5VS 7X7 88E 8FE 8FG 8FI 8FJ AADQD AAHBH AAYXX ABDBF ABUWG ACUHS ADBBV ADMLS AENEX AFFHD AFKRA AFZYC ALMA_UNASSIGNED_HOLDINGS BENPR BPHCQ BVXVI CCPQU CITATION CS3 D1I DU5 E3Z EBD ESX F5P FYUFA GROUPED_DOAJ GX1 HH5 HMCUK HYE KQ8 L6V M1P M48 MODMG M~E OK1 OVT P2P P62 PHGZM PHGZT PIMPY PJZUB PPXIY PQQKQ PROAC PSQYO RNS RPM TUS UKHRP XSB ~8M ALIPV CGR CUY CVF ECM EIF NPM 7X8 PUEGO 5PM |
| ID | FETCH-LOGICAL-c441t-ceef20362ba39d7f15e2cdc7c5de9b29fd7ec9c1eb3b4ec8b79ca440249d10bb3 |
| IEDL.DBID | DOA |
| ISICitedReferencesCount | 5 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000553180100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1424-8220 |
| IngestDate | Fri Oct 03 12:52:35 EDT 2025 Tue Nov 04 01:59:25 EST 2025 Fri Sep 05 09:10:36 EDT 2025 Thu Apr 03 07:04:25 EDT 2025 Sat Nov 29 07:12:49 EST 2025 Tue Nov 18 22:24:16 EST 2025 |
| IsDoiOpenAccess | true |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 13 |
| Keywords | inertial sensors boosting machine learning human activity recognition bagging ensemble methods |
| Language | English |
| License | Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). |
| LinkModel | DirectLink |
| MergedId | FETCHMERGED-LOGICAL-c441t-ceef20362ba39d7f15e2cdc7c5de9b29fd7ec9c1eb3b4ec8b79ca440249d10bb3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 This paper is an extended version of our paper published in Scheurer, S.; Tedesco, S.; Brown, K.N.; O’Flynn, B. Subject-dependent and-independent human activity recognition with person-specific and-independent models. In Proceedings of the 6th international Workshop on Sensor-based Activity Recognition and Interaction, Rostock, Germany, 16–17 September 2019; pp. 1–7. |
| ORCID | 0000-0002-5522-2597 0000-0002-2163-2168 0000-0002-7752-2240 |
| OpenAccessLink | https://doaj.org/article/ca3fd6fbdc394247915d8a8df36cc84b |
| PMID | 32610614 |
| PQID | 2419709932 |
| PQPubID | 23479 |
| ParticipantIDs | doaj_primary_oai_doaj_org_article_ca3fd6fbdc394247915d8a8df36cc84b pubmedcentral_primary_oai_pubmedcentral_nih_gov_7374316 proquest_miscellaneous_2419709932 pubmed_primary_32610614 crossref_citationtrail_10_3390_s20133647 crossref_primary_10_3390_s20133647 |
| PublicationCentury | 2000 |
| PublicationDate | 20200629 |
| PublicationDateYYYYMMDD | 2020-06-29 |
| PublicationDate_xml | – month: 6 year: 2020 text: 20200629 day: 29 |
| PublicationDecade | 2020 |
| PublicationPlace | Switzerland |
| PublicationPlace_xml | – name: Switzerland |
| PublicationTitle | Sensors (Basel, Switzerland) |
| PublicationTitleAlternate | Sensors (Basel) |
| PublicationYear | 2020 |
| Publisher | MDPI MDPI AG |
| Publisher_xml | – name: MDPI – name: MDPI AG |
| References | ref_14 ref_13 ref_34 ref_11 ref_10 ref_32 ref_31 ref_30 Pedregosa (ref_28) 2011; 12 ref_19 Colbert (ref_26) 2011; 13 ref_18 ref_17 ref_16 Catal (ref_12) 2015; 37 ref_25 ref_24 ref_23 ref_22 He (ref_15) 2015; 37 ref_21 ref_20 ref_1 ref_3 ref_27 Wang (ref_2) 2019; 119 ref_9 Karantonis (ref_33) 2006; 10 ref_8 ref_5 Tange (ref_29) 2011; 36 ref_4 ref_7 ref_6 |
| References_xml | – ident: ref_7 – ident: ref_21 doi: 10.1007/978-3-540-24646-6_1 – ident: ref_27 doi: 10.25080/Majora-92bf1922-00a – ident: ref_30 – ident: ref_23 doi: 10.1145/2499621 – ident: ref_32 – ident: ref_1 doi: 10.1145/3361684.3361689 – ident: ref_16 doi: 10.1109/PERCOM.2016.7456521 – ident: ref_11 – ident: ref_19 doi: 10.1109/HealthCom.2018.8531148 – ident: ref_5 doi: 10.1109/ICMLA.2015.48 – ident: ref_13 doi: 10.1109/SMC.2015.263 – volume: 10 start-page: 156 year: 2006 ident: ref_33 article-title: Implementation of a real-time human movement classifier using a triaxial accelerometer for ambulatory monitoring publication-title: IEEE Trans. Inf. Technol. Biomed. doi: 10.1109/TITB.2005.856864 – ident: ref_3 doi: 10.1016/j.patrec.2012.12.014 – volume: 13 start-page: 22 year: 2011 ident: ref_26 article-title: The NumPy Array: A Structure for Efficient Numerical Computation publication-title: Comput. Sci. Eng. doi: 10.1109/MCSE.2011.37 – ident: ref_8 doi: 10.3390/s140610146 – ident: ref_31 doi: 10.18637/jss.v067.i01 – ident: ref_14 doi: 10.1109/IJCNN.2019.8852299 – ident: ref_25 – ident: ref_4 – volume: 12 start-page: 2825 year: 2011 ident: ref_28 article-title: Scikit-learn: Machine Learning in Python publication-title: J. Mach. Learn. Res. – volume: 36 start-page: 42 year: 2011 ident: ref_29 article-title: GNU Parallel—The Command-Line Power Tool publication-title: Login Usenix Mag. – ident: ref_34 doi: 10.1017/CBO9780511790942 – ident: ref_6 doi: 10.3390/s16010115 – ident: ref_9 doi: 10.1109/ISWC.2012.13 – ident: ref_10 – volume: 37 start-page: 1018 year: 2015 ident: ref_12 article-title: On the Use of Ensemble of Classifiers for Accelerometer-Based Activity Recognition publication-title: Appl. Soft Comput. doi: 10.1016/j.asoc.2015.01.025 – ident: ref_20 doi: 10.1109/HealthCom.2016.7749439 – volume: 37 start-page: 1904 year: 2015 ident: ref_15 article-title: Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition publication-title: Trans. Pattern Anal. Mach. Intell. doi: 10.1109/TPAMI.2015.2389824 – volume: 119 start-page: 3 year: 2019 ident: ref_2 article-title: Deep learning for sensor-based activity recognition: A survey publication-title: Pattern Recognit. Lett. doi: 10.1016/j.patrec.2018.02.010 – ident: ref_24 doi: 10.1109/BSN.2017.7935994 – ident: ref_17 – ident: ref_22 – ident: ref_18 doi: 10.3390/s140610691 |
| SSID | ssj0023338 |
| Score | 2.3427765 |
| Snippet | The distinction between subject-dependent and subject-independent performance is ubiquitous in the human activity recognition (HAR) literature. We assess... |
| SourceID | doaj pubmedcentral proquest pubmed crossref |
| SourceType | Open Website Open Access Repository Aggregation Database Index Database Enrichment Source |
| StartPage | 3647 |
| SubjectTerms | Algorithms bagging boosting ensemble methods Human Activities human activity recognition Humans inertial sensors Machine Learning |
| Title | Comparing Person-Specific and Independent Models on Subject-Dependent and Independent Human Activity Recognition Performance |
| URI | https://www.ncbi.nlm.nih.gov/pubmed/32610614 https://www.proquest.com/docview/2419709932 https://pubmed.ncbi.nlm.nih.gov/PMC7374316 https://doaj.org/article/ca3fd6fbdc394247915d8a8df36cc84b |
| Volume | 20 |
| WOSCitedRecordID | wos000553180100001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVAON databaseName: Directory of Open Access Journals (DOAJ) customDbUrl: eissn: 1424-8220 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0023338 issn: 1424-8220 databaseCode: DOA dateStart: 20010101 isFulltext: true titleUrlDefault: https://www.doaj.org/ providerName: Directory of Open Access Journals – providerCode: PRVHPJ databaseName: ROAD: Directory of Open Access Scholarly Resources customDbUrl: eissn: 1424-8220 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0023338 issn: 1424-8220 databaseCode: M~E dateStart: 20010101 isFulltext: true titleUrlDefault: https://road.issn.org providerName: ISSN International Centre – providerCode: PRVPQU databaseName: ProQuest Central customDbUrl: eissn: 1424-8220 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0023338 issn: 1424-8220 databaseCode: BENPR dateStart: 20010101 isFulltext: true titleUrlDefault: https://www.proquest.com/central providerName: ProQuest – providerCode: PRVPQU databaseName: ProQuest Health & Medical Collection customDbUrl: eissn: 1424-8220 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0023338 issn: 1424-8220 databaseCode: 7X7 dateStart: 20010101 isFulltext: true titleUrlDefault: https://search.proquest.com/healthcomplete providerName: ProQuest – providerCode: PRVPQU databaseName: ProQuest Publicly Available Content Database customDbUrl: eissn: 1424-8220 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0023338 issn: 1424-8220 databaseCode: PIMPY dateStart: 20010101 isFulltext: true titleUrlDefault: http://search.proquest.com/publiccontent providerName: ProQuest |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV1Nb9QwEB1B4QAHxDehsDKIAxermziJ42M_tqKHrqIKpOUU2WMbVqqyqLvtqeK3M3ay6W5VqRcuOdgTxfKM7Te28x7AlzEWlbZa8NJZx3PUnptcFrwqDQb-c1sIH8Um5HRazWaq3pD6CnfCOnrgruP2UAtvS28sCpVnuVRpYStdWS9KxCo3YfYdS7VOpvpUS1Dm1fEICUrq95a0zInAlL61-kSS_ruQ5e0LkhsrzvFzeNZDRbbfNfEFPHDtS3i6QSD4Cq4POxnB9herI3TmUU_ez5Hp1rKTQeN2xYLo2fmSLVpGc0XYfOFHQ91t27izz_axU5ZgZ-tLRvRyffOjwWv4cTz5fviN93oKHAn0rDithz6cO2ZGC2WlTwuXoUWJhXXKZMpb6VBhSvm1yR1WRirUeR5IBW06Nka8gZ120bp3wLQmD2epQwrE3MtSZ5UXjsY_4TOrPCbwdd3PDfZk40Hz4ryhpCO4pBlcksDnwfRPx7Bxl9FBcNZgEEixYwGFStOHSnNfqCTwae3qhgZROBnRrVtcLhuCMUoSVhZZAm871w-fInwb0-YE5FZQbLVlu6ad_45E3VIEfFa-_x-N34UnWUj1xyXP1AfYWV1cuo_wGK9W8-XFCB7KmYzPagSPDibT-mwURwQ9T_9OqKw-Oa1__gPc_hgi |
| linkProvider | Directory of Open Access Journals |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Comparing+Person-Specific+and+Independent+Models+on+Subject-Dependent+and+Independent+Human+Activity+Recognition+Performance&rft.jtitle=Sensors+%28Basel%2C+Switzerland%29&rft.au=Scheurer%2C+Sebastian&rft.au=Tedesco%2C+Salvatore&rft.au=O%27Flynn%2C+Brendan&rft.au=Brown%2C+Kenneth+N&rft.date=2020-06-29&rft.eissn=1424-8220&rft.volume=20&rft.issue=13&rft_id=info:doi/10.3390%2Fs20133647&rft_id=info%3Apmid%2F32610614&rft.externalDocID=32610614 |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1424-8220&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1424-8220&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1424-8220&client=summon |