KIT | KIT-Bibliothek | Impressum | Datenschutz

Improving Human Activity Recognition Models by Learnable Sparse Wavelet Layer

Zhao, Haibin ORCID iD icon 1; Zhou, Yexu 1; Riedel, Till ORCID iD icon 1; Hefenbrock, Michael 1; Beigl, Michael 1
1 Institut für Telematik (TM), Karlsruher Institut für Technologie (KIT)

Abstract (englisch):

Modern machine learning algorithms for human activity recognition based on artificial neural networks often require a large amount of labelled training data to generalize between human subjects and training contexts. Large degrees of freedom make them susceptible to overfitting and often computationally intensive to implement on portable hardware. In this work, we introduce wavelet-based learnable filters as a feature extraction layer that greatly improves the generalization capability of the detector model. Our evaluations on six benchmark datasets show significant improvements in macro $F_1$ score when our wavelet-based learnable filter layer is prepended to three state-of-the-art human activity recognition models. As a side effect, in many cases we could drastically reduce the required model size to achieve competitive performance on the benchmark dataset, which is an important requirement for use in wearable computing.

Postprint §
DOI: 10.5445/IR/1000150214/post
Veröffentlicht am 12.09.2023
Preprint §
DOI: 10.5445/IR/1000150214
Veröffentlicht am 05.09.2022
DOI: 10.1145/3544794.3558461
Zitationen: 1
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Telematik (TM)
Publikationstyp Proceedingsbeitrag
Publikationsjahr 2022
Sprache Englisch
Identifikator ISBN: 978-1-4503-9424-6
KITopen-ID: 1000150214
Erschienen in International Symposium on Wearable Computers (ISWC'22) , Atlanta, GA and Cambridge, UK, September 11-15, 2022
Veranstaltung International Symposium on Wearable Computers (ISWC 2022), Cambridge, Vereinigtes Königreich, 11.09.2022 – 15.09.2022
Verlag Association for Computing Machinery (ACM)
Seiten 84-88
Projektinformation JuBot (ZEISS-STFG, JuBot)
CC-KING (WM_BW, 3-4332.62-FhG/38)
SDI-C (BMBF, 01IS19030A)
Vorab online veröffentlicht am 11.09.2022
Schlagwörter human activity recognition, learnable filters, wavelet analysis
Nachgewiesen in Dimensions
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page