Digging deeper: towards a better understanding of transfer learning for human activity recognition
The DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index.
Abstract
Transfer Learning is becoming increasingly important to the Human Activity Recognition community, as it enables algorithms to reuse what has already been learned from models. It promises shortened training times and increased classification results for new datasets and activity classes. However, the question of what exactly is transferred is not dealt with in detail in many of the recent publications, and it is furthermore often difficult to reproduce the presented results. Therefore we would like to contribute with this paper to the understanding of transfer learning for sensor-based human activity recognition. In our experiment use weight transfer to transfer models between two datasets, as well as between sensors from the same dataset. As source- and target- datasets PAMAP2 and Skoda Mini Checkpoint are used. The utilized network architecture is based on a DeepConvLSTM. The result of our investigation shows that transfer learning has to be considered in a very differentiated way, since the desired positive effects by applying the method depend very much on the data and also on the architecture used.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“the research community has explored many different avenues, ranging from, for example, data augmentation [10,11], transfer learning [12], to self-supervised learning techniques [13].”
From Complex Deep Neural Networks from Large Scale Virtual IMU Data for Effective Human Activity Recognition Using Wearables (2021) · cites this paper as [12]
Citations
Cites
Learning Transferable Features with Deep Adaptation Networks
Long, M., Cao, Y., Wang, J., & Jordan, M. I. (2015) · arXiv · doi:10.48550/arxiv.1502.02791 · Cited by 4,505 · cited 1×
“used in [14] and [15], to transfer the pretrained model.”
MethodActivity Recognition from User-Annotated Acceleration Data
Bao, L., & Intille, S. S. (2004) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-24646-6_1 · Cited by 3,147 · cited 2×
“Where traditional methods reach their limits, either because of the way classification algorithms work or because of the required feature engineering steps of handmade features [1], neural networks manage to close the gap and outperform classical ML approaches regularly [21].”
BackgroundDeep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Ordóñez, F., & Roggen, D. (2016) · Sensors · doi:10.3390/s16010115 · Cited by 2,761
Deep learning for sensor-based activity recognition: A survey
Wang, J., Chen, Y., Hao, S., Peng, X., & Hu, L. (2019) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.02.010 · Cited by 2,014 · cited 1×
“Where traditional methods reach their limits, either because of the way classification algorithms work or because of the required feature engineering steps of handmade features [1], neural networks manage to close the gap and outperform classical ML approaches regularly [21].”
BackgroundIntroducing a New Benchmarked Dataset for Activity Monitoring
Reiss, A., & Stricker, D. (2012) · 2012 16th International Symposium on Wearable Computers · doi:10.1109/iswc.2012.13 · Cited by 1,456 · cited 1×
“By using an architecture called MultiResNet [5], that transfers the data into frequency domain and uses residual blocks they achieved promising results when transferring from Skoda Mini Checkpoint [23] to OPPORTUNITY [18], PAMAP2 [17] or JSI-FOS [4].”
MethodCollecting complex activity datasets in highly rich networked sensor environments
Roggen, D., Calatroni, A., Rossi, M., Holleczek, T., Forster, K., Troster, G., Lukowicz, P., Bannach, D., Pirkl, G., Ferscha, A., et al. (2010) · 2010 Seventh International Conference on Networked Sensing Systems (INSS) · doi:10.1109/inss.2010.5573462 · Cited by 788
Deep, Convolutional, and Recurrent Models for Human Activity Recognition Using Wearables
Hammerla, N. Y., Halloran, S., & Ploetz, T. (2016) · arXiv · doi:10.48550/arxiv.1604.08880 · Cited by 606 · cited 1×
“PAMAP2, on the other hand, already proved in previous publications, for example [16] or [8], to be suitable for use with neural networks.”
BackgroundPD Disease State Assessment in Naturalistic Environments Using Deep Learning
Hammerla, N., Fisher, J., Andras, P., Rochester, L., Walker, R., & Ploetz, T. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9484 · Cited by 179 · cited 2×
“[7] shows, that the most important advantage of a Deep Neural Network against traditional Machine Learning approaches, e.”
BackgroundMethodMorales, F. J. O., & Roggen, D. (2016) · Proceedings of the 2016 ACM International Symposium on Wearable Computers · doi:10.1145/2971763.2971764 · Cited by 136 · cited 5×
“All methods follow the weight transfer method, e.g. used in [14] and [15], to transfer the pretrained model.”
BackgroundMethodResultInfluentialA-Wristocracy: Deep learning on wrist-worn sensing for recognition of user complex activities
Vepakomma, P., De, D., Das, S. K., & Bhansali, S. (2015) · 2015 IEEE 12th International Conference on Wearable and Implantable Body Sensor Networks (BSN) · doi:10.1109/bsn.2015.7299406 · Cited by 136 · cited 1×
“While in the beginning it was still tried to use handmade features as input data [20], since [7] at the latest, the data are now treated directly as time-discrete signals and thus can be used as input for the neural networks.”
MethodSensor Placement Variations in Wearable Activity Recognition
Kunze, K., & Lukowicz, P. (2014) · IEEE Pervasive Computing · doi:10.1109/mprv.2014.73 · Cited by 111 · cited 1×
“[ 15] showed a setup that we build up on and expanded with tests that artificially mapped the sensors placement and orientation to each other, according to the results of [12] and [24].”
BackgroundIMU-to-Segment Assignment and Orientation Alignment for the Lower Body Using Deep Learning
Zimmermann, T., Taetz, B., & Bleser, G. (2018) · Sensors · doi:10.3390/s18010302 · Cited by 99 · cited 1×
“[ 15] showed a setup that we build up on and expanded with tests that artificially mapped the sensors placement and orientation to each other, according to the results of [12] and [24].”
BackgroundGjoreski, M., Janko, V., Slapničar, G., Mlakar, M., Reščič, N., Bizjak, J., Drobnič, V., Marinko, M., Mlakar, N., Luštrek, M., et al. (2020) · Information Fusion · doi:10.1016/j.inffus.2020.04.004 · Cited by 96 · cited 1×
“By using an architecture called MultiResNet [5], that transfers the data into frequency domain and uses residual blocks they achieved promising results when transferring from Skoda Mini Checkpoint [23] to OPPORTUNITY [18], PAMAP2 [17] or JSI-FOS [4].”
MethodCross-domain activity recognition via transfer learning
Hu, D. H., Zheng, V. W., & Yang, Q. (2010) · Pervasive and Mobile Computing · doi:10.1016/j.pmcj.2010.11.005 · Cited by 93
Deep Transfer Learning for Time Series Data Based on Sensor Modality Classification
Li, F., Shirahama, K., Nisar, M. A., Huang, X., & Grzegorzek, M. (2020) · Sensors · doi:10.3390/s20154271 · Cited by 74 · cited 1×
“[11], [22], [3], [2], [15] or [13], especially in the last two years [9].”
BackgroundHernandez-Cruz, N., Lundström, J., Favela, J., McChesney, I., & Arnrich, B. (2020) · SN Computer Science · doi:10.1007/s42979-020-0070-4 · Cited by 70
Context-based ensemble method for human energy expenditure estimation
Gjoreski, H., Kaluža, B., Gams, M., Milić, R., & Luštrek, M. (2015) · Applied Soft Computing · doi:10.1016/j.asoc.2015.05.001 · Cited by 62 · cited 1×
“By using an architecture called MultiResNet [5], that transfers the data into frequency domain and uses residual blocks they achieved promising results when transferring from Skoda Mini Checkpoint [23] to OPPORTUNITY [18], PAMAP2 [17] or JSI-FOS [4].”
MethodMcfly: Automated deep learning on time series
van Kuppevelt, D. E., Meijer, C., Huber, F., van der Ploeg, A., Georgievska, S., & van Hees, V. T. (2020) · SoftwareX · doi:10.1016/j.softx.2020.100548 · Cited by 55 · cited 1×
“A variation of this architecture inspired by the McFly-Project for Deep Learning [19] is used in our experiments.”
MethodZappi, P., Roggen, D., Farella, E., Tröster, G., & Benini, L. (2012) · TECS · doi:10.1145/2345770.2345781 · Cited by 48 · cited 1×
“By using an architecture called MultiResNet [5], that transfers the data into frequency domain and uses residual blocks they achieved promising results when transferring from Skoda Mini Checkpoint [23] to OPPORTUNITY [18], PAMAP2 [17] or JSI-FOS [4].”
MethodUsing Wrist-Worn Activity Recognition for Basketball Game Analysis
Hölzemann, A., & Laerhoven, K. V. (2018) · International Workshop on Sensor-based Activity Recognition and Interaction (iWOAR) · doi:10.1145/3266157.3266217 · Cited by 38 · cited 1×
“[10], is that the traditional approaches work with hand crafted features [1], whereas neural networks learn their features dynamically through optimizing their connecting weights.”
BackgroundActive transfer learning for activity recognition
Diethe, T., Twomey, N., & Flach, P. A. (2016) · Bristol Research (University of Bristol) · Cited by 28
Cross-dataset deep transfer learning for activity recognition
Gjoreski, M., Kalabakov, S., Luštrek, M., Gams, M., & Gjoreski, H. (2019) · Adjunct Proceedings of the 2019 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2019 ACM International Symposium on Wearable Computers · doi:10.1145/3341162.3344865 · Cited by 28 · cited 2×
“Two publications that influenced our choice of datasets are [6] and [15].”
BackgroundChen, Y., Gu, Y., Jiang, X., & Wang, J. (2016) · ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp) · doi:10.1145/2968219.2971453 · Cited by 12 · cited 1×
“The number of published papers in this discipline has increased rapidly, e.g. [ 11], [22], [3], [2], [15] or [13], especially in the last two years [9].”
BackgroundTransfer Learning on High Variety Domains for Activity Recognition
Ying, J. J.-C., Lin, B.-H., Tseng, V. S. M., & Hsieh, S. (2015) · ASE BigData & SocialInformatics · doi:10.1145/2818869.2818890 · Cited by 11 · cited 1×
“[11], [22], [3], [2], [15] or [13], especially in the last two years [9].”
Background
Research works printed but not linked: 3
- 10.5555/2886521.2886562 · doi:10.5555/2886521.2886562
- Long Mingsheng, 2015, International conference on machine learning. 97--105
- Jia-Ching Ying Josh, 2015, Proceedings of the ASE BigData & SocialInformatics 2015 (ASE BD&SI '15)
Cited by
Kumar, P., Chauhan, S., & Awasthi, L. K. (2024) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-023-09986-x · Cited by 206
COCOA: Cross Modality Contrastive Learning for Sensor Data
Deldari, S., Xue, H., Saeed, A., Smith, D. V., & Salim, F. D. (2022) · Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies · doi:10.1145/3550316 · Cited by 92
Hoelzemann, A., Romero, J., Bock, M., Laerhoven, K. V., & Lv, Q. (2023) · Sensors · doi:10.3390/s23135879 · Cited by 61
Self-citationHuman Activity Recognition Data Analysis: History, Evolutions, and New Trends
Ariza-Colpas, P. P., Vicario, E., Oviedo-Carrascal, A. I., Butt Aziz, S., Piñeres-Melo, M. A., Quintero-Linero, A., & Patara, F. (2022) · Sensors · doi:10.3390/s22093401 · Cited by 52
Bento, N., Rebelo, J., Barandas, M., Carreiro, A. V., Campagner, A., Cabitza, F., & Gamboa, H. F. S. (2022) · Sensors · doi:10.3390/s22197324 · Cited by 51
“[7], significant drops in performance were reported when testing on different positions and different datasets, which were then mitigated by the use of transfer learning techniques.”
BackgroundMethod
Metadata
- Authors
- Hoelzemann, A., & Van Laerhoven, K.
- Affiliations
- University of Siegen (DE)
- University of Siegen
- Venue
- Proceedings of the 2020 International Symposium on Wearable Computers
- Publisher
- ACM
- Published
- 2020-09-04
- Pages
- 50-54
- OpenAlex
- W3083016735
- Cited by
- 20
- References
- 24
- Citations by year
- 2025: 3 · 2024: 3 · 2023: 5 · 2022: 5 · 2021: 4
- FWCI
- 1.08
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- International Symposium on Wearable Computers (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
- Keywords
- transfer learning, human activity recognition, sensor-based activity recognition, weight transfer, PAMAP2, cross-dataset transfer, reproducibility, convolutional LSTM
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.