A Fine-Tuning Based Approach for Daily Activity Recognition between Smart Homes
Source abstract
Abstract
Daily activity recognition between different smart home environments faces some challenges, such as an insufficient amount of data and differences in data distribution. However, a deep network requires a large amount of labeled data for training. Additionally, inconsistent data distribution can lead to over-fitting of network learning. Additionally, the time cost of training the network from scratch is too high. In order to solve the above problems, this paper proposes a fine-tuning method suitable for daily activity recognition, which is the first application of this method in our field. Firstly, we unify the sensor space and activity space to reduce the variability in heterogeneous environments. Then, the Word2Vec algorithm is used to transform the activity samples into digital vectors recognizable by the network. Finally, the deep network is fine-tuned to transfer knowledge and complete the recognition task. Additionally, we try to train the network on public datasets. The results show that the network trained on a small dataset also has good transferability. It effectively improves the recognition accuracy and reduces the time cost and heavy data annotation.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Context-Aware Activity Recognition Systems · Human Pose and Action Recognition · Human Mobility and Location-Based Analysis
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
School of Information Science & Technology, Dalian Maritime University, Dalian 116026, China
References (36)
Show source-provided references
1. Interaction-Feedback Network for Multi-Task Daily Activity Forecasting10.1016/j.eswa.2023.119602 ↗
2. Kasteren, V., Englebienne, G., and Krose, B.J.A. (2008, January 7–9). Recognizing activities in multiple contexts using transfer learning. Proceedings of the AAAI Symposium, Arlington, VA, USA.
3. A survey on transfer learning10.1109/TKDE.2009.191 ↗
4. Transfer learning for activity recognition: A survey10.1007/s10115-013-0665-3 ↗
5. Optimal search mapping among sensors in heterogeneous smart homes10.3934/mbe.2023090 ↗
6. Fawaz, H.I., Forestier, G., Weber, J., Idoumghar, L., and Muller, P.-A. (2018, January 10–13). Transfer learning for time series classification. Proceedings of the IEEE International Conference on Big Data, Seattle, WA, USA.
7. Chen, Y., Keogh, E., Hu, B., Yeh, C.C.M., Zhu, Y., Gharghabi, S., Ratanamahatana, C.A., and Keogh, E. (2018, October 15). The UCR Time Series Classification Archive. Available online: https://www.cs.ucr.edu/~eamonn/time_series_data_2018.
8. CASAS: A smart home in a box10.1109/MC.2012.328 ↗
9. XLearn: Learning activity labels across heterogeneous datasets10.1145/3368272 ↗
10. Transfer learning across feature-rich heterogeneous feature spaces via feature-space remapping (FSR)10.1145/2629528 ↗
11. Cross-environment activity recognition using word embeddings for sensor and activity representation10.1016/j.neucom.2020.08.044 ↗
12. Chiang, Y.T., and Hsu, Y.J. (2012, January 4–7). Knowledge transfer in activity recognition using sensor profile. Proceedings of the International Conference on Ubiquitous Intelligence and Computing and International Conference on Autonomic and Trusted Computing, Washington, DC, USA.10.1109/UIC-ATC.2012.78 ↗
13. Hu, D.H., and Yang, Q. (2011, January 16–22). Transfer learning for activity recognition via sensor mapping. Proceedings of the Twenty-Second International Joint Conference on Artificial Intelligence, Barcelona, Spain.
14. Rashidi, P., and Cook, D.J. (2010). Activity Recognition Based on Home to Home Transfer Learning, AAAI Press.
15. Cross-domain activity recognition via transfer learning10.1016/j.pmcj.2010.11.005 ↗
16. How transferable are features in deep neural networks?
17. Imagenet classification with deep convolutional neural networks
18. Tzeng, E., Hoffman, J., Zhang, N., Saenko, K., and Darrell, T. (2014). Deep domain confusion: Maximizing for domain invariance. arXiv.
19. Integrating structured biological data by kernel maximum mean discrepancy10.1093/bioinformatics/btl242 ↗
20. Long, M., Cao, Y., Wang, J., and Jordan, M. (2015, January 6–11). Learning transferable features with deep adaptation networks. Proceedings of the International Conference on Machine Learning, Lille, France.
21. Optimal kernel choice for largescale two-sample tests
22. Zhuo, J., Wang, S., Zhang, W., and Huang, Q. (2017, January 23–27). Deep unsupervised convolutional domain adaptation. Proceedings of the 2017 ACM on Multimedia Conference, Mountain View, CA, USA.10.1145/3123266.3123292 ↗
23. Deep learning of transferable representation for scalable domain adaptation10.1109/TKDE.2016.2554549 ↗
24. Sun, B., and Saenko, K. (2016, January 11–14). Deep coral: Correlation Alignment for Deep Domain Adaptation. Proceedings of the European Conference on Computer Vision, Amsterdam, The Netherlands.10.1007/978-3-319-49409-8_35 ↗
25. Wei, P., Ke, Y., and Goh, C.K. (2016, January 9–15). Deep nonlinear feature coding for unsupervised domain adaptation. Proceedings of the International Joint Conference on Artificial Intelligence, New York, NY, USA.
26. Zhuang, F., Cheng, X., Luo, P., Pan, S.J., and He, Q. (2015, January 25–31). Supervised Representation Learning: Transfer Learning with Deep Autoencoders. Proceedings of the International Joint Conference on Artificial Intelligence, Buenos Aires, Argentina.
27. Label efficient learning of transferable representations acrosss domains and tasks
28. Summarizing a set of time series by averaging: From steiner sequence to compact multiple alignment10.1016/j.tcs.2011.09.029 ↗
29. Performance of End-to-end Model Based on Convolutional LSTM for Human Activity Recognition
30. Singh, D., Merdivan, E., Hanke, S., Geist, M., and Holzinger, A. (2015, January 24–26). Convolutional and Recurrent Neural Networks for Activity Recognition in Smart Environment. Proceedings of the Banff-International-Research-Station (BIRS) Workshop, Banff, AB, Canada.
31. A sequential deep learning application for recognising human activities in smart homes10.1016/j.neucom.2018.10.104 ↗
32. Adapted Long Short-Term Memory (LSTM) for Concurrent Human Activity Recognition
33. Forbes, G., Massie, S., Craw, S., Fraser, L., and Hamilton, G. (2020, January 19–24). Representing Temporal Dependencies in Smart Home Activity Recognition for Health Monitoring. Proceedings of the International Joint Conference on Neural Networks, Glasgow, UK.10.1109/IJCNN48605.2020.9207480 ↗
34. Mikolov, T., Chen, K., Corrado, G., and Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv.
35. Anguita, D., Ghio, A., Oneto, L., Parra, X., and Reyes-Ortiz, J.L. (2013, January 24–26). A public domain dataset for human activity recognition using smartphones. Proceedings of the European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Bruges, Belgium.
36. Understanding the difficulty of training deep feed forward neural networks