人口老齡化給整個社會帶來很多問題，如醫療資源及護理人員的嚴重短缺、護理費用高昂等。考慮到多數老人更喜歡在家中安享晚年，同時結合老齡化產業現現狀，居家養老成為大多數政府主推的養老模式。而居家照護也成為了一個熱門研究議題，受到了廣泛的關注。
行為識別在支持老人居家養老中起著重要的作用。現有研究提出的行為識別演算法，多是通過從感測器資料中提取特徵或模式來識別老年人的行為。然而，這些方法大多基於帶有標記的感測器資料，採用概率模型或監督學習的方法來識別行為。本研究針對智慧家居中的獨居老人，基於無標記的感測器資料，提出了一種基於無監督學習的行為識別算法(BIA)。本研究基於對老人行為的觀察，提出了三個老人行為的特徵，即事件順序相似性、時間長度相似性和時間間隔相似性。基於這些行為觀察的特徵，定義了兩種行為屬性，即事件位移和長條圖形狀相似度。根據這些特性，提出了無監督學習的行為識別算法（BIA）。最後，實驗結果表明，該方法在行為識別精度和召回率方面優於現有的無監督機器學習機制。
老年人日常行為的異常檢測也是居家照護中的一個重要議題。現有的異常檢測研究多是基於某些生物醫學參數或某些特定行為的明顯異常來評估老年人的身體健康狀態。然而，很少有研究討論不同行為組合的隱性異常，這種隱性異常可以用來評估老年人的認知和身體健康，但不能基於感測器資料直接識別。因此，本研究提出一種隱式不規則檢測(IIRD)機制，旨在基於日常行為應用無監督學習算法來檢測老人行為為的隱式不規則性。本研究提出的IIRD機制能夠識別日常行為之間的距離和相似性，這是區分老年人日常行為的規律性和不規則性、檢測老年人健康狀況隱性不規則性的重要特徵。實驗結果表明，該方法在檢測準確率和Recall方面優於現有的無監督機器學習機制。
由於IIRD只輸出二值檢測結果，因此本研究進一步提出了一種基於特徵的隱式不規則性檢測機制(FIID)，該機制利用無監督學習提取規則性特徵，輸出隱式不規則性發生的概率。該方法將滿足時間規律性性和頻繁發生規律性性的規則行為作為識別為日常行為的規則特徵。這些特徵可以構造一個多維特徵空間來計算日常健康狀況的隱式不規則概率。實驗結果表明，本研究所提出的FIID在精度、Recall和F-measure方面都優於現有的隱式不規則機制。

Advances in wireless sensor networks and increasing Internet-of-Things devices give great opportunities for smart homecare of the elderly. Smart homecare has been a promising issue and received much attention recently. 
Behavior identification plays an important role in supporting homecare for the elderly living alone. In literature, plenty of algorithms have been designed to identify behaviors of the elderly by learning features or extracting patterns from sensor data. However, most of them adopted probabilistic models or supervised learning to identify behaviors based on labeled sensor data. This study proposes a behavior identification algorithm (BIA) using unsupervised learning based on unlabeled sensor data for the elderly living alone in smart home. This study presents the observation of elder behaviors with three features: Event Order, Time Length Similarity and Time Interval Similarity features. Based on these features of behavior observations, two properties of behaviors, including the Event Shift and Histogram Shape Similarity properties, are presented. According to these properties, the BIA is developed. Finally, performance results show that the BIA outperforms the existing unsupervised machine learning mechanisms in terms of the behavior identification precision and recall. 
The irregularity detection of daily behaviors for the elderly also is an important issue in homecare. Plenty of mechanisms have been developed to detect the health condition of the elderly based on the explicit irregularity of several biomedical parameters or some specific behaviors. However, few research works focus on detecting the implicit irregularity involving the combination of diverse behaviors, which can assess the cognitive and physical wellbeing of elders but cannot be directly identified based on sensor data. This study proposes an Implicit IRregularity Detection (IIRD)mechanism that aims to detect the implicit irregularity by developing the unsupervised learning algorithm based on daily behaviors. The IIRD mechanism identifies the distance and similarity between daily behaviors, which are important features to distinguish the regular and irregular daily behaviors and detect the implicit irregularity of elderly health condition. Performance results show that the IIRD outperforms the existing unsupervised machine-learning mechanisms in terms of the detection accuracy and irregularity recall. 
Since the IIRD simply outputs the binary detection results, this study proposes a feature-based implicit irregularity detection mechanism (FIID) which extracts the regularity features using unsupervised learning and outputs the probability of implicit irregularity. The FIID identifies the regular behaviors which satisfy the time-regular and happen-frequently properties as the regularity features of daily behaviors. These features then construct a multidimensional feature space to calculate the implicit irregularity probability of the daily health condition. Performance results show that the FIID outperforms the existing implicit irregularity mechanism in terms of precision, recall as well as F-measure.

Outline
Outline	V
List of Figures	VIII
List of Tables	X
Chapter 1. Introduction	1
1.1 Background	1
1.2 Research Goals	3
1.3 Organization of the Thesis	4
Chapter 2. Related Work	5
2.1 Behaviors Identification	5
2.1.1 Vision-based Algorithms	5
2.1.2 Wearable-based Algorithms	5
2.1.3 Unobtrusive Sensor-based Algorithms	5
2.2 Irregular Detection	7
2.2.1 Explicit Irregularity Detection	7
2.2.2 Implicit Irregularity Detection	9
Chapter 3. The Behavior Identification Algorithm	11
3.1 Assumptions and Problem Statement	12
3.1.1 Network Environment	12
3.1.2 Problem Statement	13
3.2 The Properties of Behaviors	15
3.2.1 Observation on the Elder Behaviors	16
3.2.2 Histogram of Sensor Events	18
3.3 The Design of BIA	21
3.3.1 Sensor Event Grouping Phase	21
3.3.2 Histogram Similarity Grouping Phase	23
3.3.3 Behavior Matching Phase	27
3.4 Simulation	28
3.4.1 Data Description	28
3.4.2 Experimental Results	30
3.5 Summary	35
Chapter 4. The Implicit Irregularity Detection Mechanism (IIRD)	37
4.1 Assumptions and Problem Statement	38
4.1.1 Network Environment	38
4.1.2 Problem Statement	39
4.1.3 Assumptions and Constraints	41
4.2 Distance and Similarity Calculations	42
4.2.1 Distance Calculation	42
4.2.2 Similarity Calculation	45
4.3 The Design of IIRD	47
4.3.1 Construction of Basic Regular Group (CBRG)	47
4.3.2 Expansion of Regular Group(ERG)	48
4.3.3 The Design of IIRD Mechanism	50
4.4 Simulation	51
4.4.1 Data Preparation	51
4.4.2 Experimental Results	55
4.5 Summary	60
Chapter 5. The Feature-based Implicit Irregularity Detection Mechanism (FIID)	62
5.1 Assumptions and Problem Statement	63
5.1.1 System Model and Assumptions	63
5.1.2 Problem Statements	63
5.2 The Design of FIID	65
5.2.1 Selection of Regular Behaviors Phase	65
5.2.2 Mapping of Daily Behavior Sequence Phase	70
5.2.3 Identification of Irregular Days Phase	73
5.3 Simulation of FIID	76
5.3.1 Data Description	77
5.3.2 Experimental Results	78
5.4 Summary	86
Chapter 6. Conclusion and Future Work	87
References	89

 
List of Figures
Fig. 3.1. The example of Similarity calculation of H_j and H_k.	20
Fig. 3.2. The procedure of partitioning event set E in SEG phase.	22
Fig. 3.3. The procedure of partitioning histogram set H in HSG phase.	24
Fig. 3.4. The merging of clusters and the construction of a hierarchical tree.	26
Fig. 3.5. The information of actual behaviors of a month	29
Fig. 3.6. The occurrence frequency of sensor events.	30
Fig. 3.7. The performance of the sby varying Eps and min_pts.	31
Fig. 3.8. The performances of some interested behaviors, in terms of recall and precision.	31
Fig. 3.9. The performance of the BIA, in terms of total recall and total precision by varying cutoff coefficient Cc and class interval τ.	33
Fig. 3.10. The performance, in terms of recall, of some interested behaviors.	33
Fig. 3.11. The cluster tree of the region bathroom	34
Fig. 3.12. Performance of three compared algorithms in terms of recall, precision, and F1 score	35
Fig. 4.1. Representation of daily behaviors.	39
Fig. 4.2. Distance calculation of two daily behaviors.	45
Fig. 4.3. Origin o_G and diameter d_G of group G.	46
Fig. 4.4. The IIRD algorithm.	51
Fig. 4.5. Example of the derivations of one regular and one irregular strings from basic regular string.	53
Fig. 4.6. The derivations regular strings which are 75 irregular strings selected from 365 days.	53
Fig. 4.7. Projecting strings onto a two-dimensional plane.	54
Fig. 4.8. False positive/false negative of four mechanisms.	55
Fig. 4.9. Detection accuracy and irregularity recall (vs number of training data).	57
Fig. 4.10. Detection accuracy and irregularity recall (vs number of detection data).	57
Fig. 4.11. Detection accuracy (vs ratio of irregular data to training data).	58
Fig. 4.12. Irregularity recall (vs ratio of irregular data to training data).	59
Fig. 4.13. Detection accuracy of four mechanisms (vs ratio ε and ratio δ).	59
Fig. 4.14. Discrimination of IIRD.	60
Fig. 5.1. An example of the R-space construction.	69
Fig. 5.2. The procedure of selection of regular behaviors phase.	70
Fig. 5.3. An example of mapping of daily behavior sequence.	73
Fig. 5.4. An example of R-space and irregularity probability calculation.	75
Fig. 5.5. The information of actual behaviors of a month.	77
Fig. 5.6. The occurrence frequency of sensor events.	77
Fig. 5.7. The performance of DBSCAN by varying Eps and min⁡_pts.	80
Fig. 5.8. The impacts of the threshold _trh.	81
Fig. 5.9. A snapshot of checking the criterion for having happen-frequently property.	81
Fig. 5.10. The snapshot of regular behaviors.	82
Fig. 5.11. The performance of FIID by varying the probability threshold Δ_trh.	83
Fig. 5.12. Performance of four compared algorithms by varying the values of parameters Min_pts and Eps.	84
Fig. 5.13. Performance of four compared algorithms by varying the data size.	85

 
List of Tables
Table 2.1. Comparison of the main characteristics of the BIA with the unobtrusive sensor-based studies	7
Table 2.2. Comparison of the main characteristics of the IIRD and FIID with the existing related works	10
Table 3.1. Sensor data format	29
Table 4.1. Simulation parameters	52
Table 4.2. Distance feature of derived strings	59

[1]	F. Zhou, J. Jiao, S. Chen and D. Zhang, "A case-driven ambient intelligence system for elderly in-home assistance applications," IEEE Trans. Syst. Man and Cybern., vol. 41, no. 2, pp. 179-189, Mar. 2011.
[2]	J. Rafferty, C. D. Nugent, J. Liu and L. Chen, "From activity recognition to intention recognition for assisted living within smart homes," IEEE Trans. On Human-Machine Systems, vol. 47, no. 3, pp. 368-379, June 2017.
[3]	C. Tsirmpas, A. Anastasiou, P. Bountris, and D. Koutsouris, "A new method for profile generation in an intemet of things environment: An application in ambient-assisted living," IEEE Internet of Things J., vol. 2, no. 6, pp. 471-478, Dec. 2015.
[4]	H. Habibzadeh, et al., "A survey of healthcare Internet of Things (HIoT): a clinical perspective," IEEE Internet Things J., vol. 7, no. 1, pp.53-71, Jan. 2020.
[5]	D. Naranjo-Hernández, L. M. Roa, J. Reina-Tosina and M. Á. Estudillo-Valderrama, "SoM: A smart sensor for human activity monitoring and assisted healthy ageing," IEEE Trans. Biomed. Eng., vol. 59, no. 11, pp. 3177-3184, Nov. 2012.
[6]	L. Chen, J. Hoey, C. D. Nugent, D. J. Cook and Z. Yu, "Sensor-based activity recognition," IEEE Trans. Syst. Man and Cybern., vol. 42, no. 6, pp. 790-808, Nov. 2012.
[7]	J. J. Liu et al., "BreathSens: A Continuous On-Bed Respiratory Monitoring System With Torso Localization Using an Unobtrusive Pressure Sensing Array," IEEE J. Biomed. Health Inform., vol. 19, no. 5, pp. 1682-1688, Sep. 2015.
[8]	T. S. Barger, D. E. Brown, and M. Alwan, "Health-status monitoring through analysis of behavioral patterns," IEEE Trans. Syst. Man and Cybern., vol. 35, no. 1, pp. 22-27, Jan. 2005.
[9]	C. Zhu, and W. Sheng, "Realtime recognition of complex human daily activities using human motion and location data," IEEE Trans. Biomed. Eng., vol. 59, no. 9, pp. 2422-2430, Sep. 2012.
[10]	Y. Chen, L. Yu, K. Ota and M. Dong, "Robust activity recognition for aging society," IEEE J. Biomed. Health Inform., vol. 22, no. 6, pp. 1754-1764, Nov. 2018.
[11]	F. Osayamwen and J.-R. Tapamo, "Deep learning class discrimination based on prior probability for human activity recognition," IEEE Access, vol. 7, pp. 14747-14746, Jan. 2019. 
[12]	Y.Yan, E. Ricci, G. Liu and N. Sebe, "Egocentric Daily Activity Recognition via Multitask Clustering," IEEE Ttrans. Image Processing, vol. 24, no. 10, pp. 2984-2995, Oct. 2015. 
[13]	M. Wang, C. Luo, B. Ni, J. Yuan, J. Wang and S. Yan, "First-person daily activity recognition with manipulated object proposals and non-linear feature fusion," IEEE Trans. Circuits and Systems for Video Technology, vol. 28, no. 10, pp. 2946-2955, Oct. 2018. 
[14]	W. Xu, Z. Miao, X.-P. Zhang and Y. Tian, "A hierarchical spatio-temporal model for human activity recognition," IEEE Trans. Multimedia, vol. 19, no. 7, pp. 1494-1509, Jul. 2017.
[15]	G. Valenza, M. Nardelli, A. Lanata`, C. Gentili, G. Bertschy, R. Paradiso and E. P. Scilingo, "Wearable monitoring for mood recognition in bipolar disorder based on history-dependent long-term heart rate variability analysis," IEEE J. Biomed. Health Inform., vol. 18, no. 5, pp. 1625-1635, Sep. 2014.
[16]	Nazábal, P. García-Moreno, A. Artés-Rodríguez, and Z. Ghahramani, "Human activity recognition by combining a small number of classifiers," IEEE J. Biomed. Health Inform., vol. 20, no. 5, pp. 1342-1351, Sep. 2016.
[17]	D. Tao, L. Jin, Y. Yuan and Y. Xue, "Ensemble manifold rank preserving for acceleration-based human activity recognition," IEEE Trans. Neural Networks And Learning Systems, vol. 27, no. 6 , pp. 1392-1404, Jun. 2016.
[18]	S. Zhang, S. I. McClean and B. W. Scotney, "Probabilistic learning from incomplete data for recognition of activities of daily living in smart homes," IEEE Trans. Information Technology in Biomedicine, vol. 16, no. 3, pp. 454-462, May 2012.
[19]	Y. Zhang, et al., "Ubiquitous WSN for healthcare: recent advances and future prospects," IEEE Internet Things J., vol. 1, no. 4, pp. 311-318, Aug. 2014.
[20]	 D. J. Cook, N. C. Krishnan and P. Rashidi, "Activity Discovery and Activity Recognition: A New Partnership," IEEE Trans. Cybern, vol. 43, no. 3, pp. 820-828, Jun. 2013.
[21]	L. Peng, L. Chen, X. Wu, H. Guo and G. Chen, “Hierarchical complex activity representation and recognition using topic model and classifierlevel fusion,” IEEE Trans. Biomedical Engineering, vol. 64, no. 6, pp. 1369-1379, Jun. 2017.
[22]	P. Gupta and T. Dallas, "Feature selection and activity recognition system using a single triaxial accelerometer," IEEE Trans. Biomedical Engineering, vol. 61, no. 6, pp. 1780-1786, Jun. 2014. 
[23]	T. Magherini, A. Fantechi, C. D. Nugent and E. Vicario, "Using temporal logic and model checking in automated recognition of human activities for ambient-assisted living," IEEE Trans. Human-Machine Systems, vol. 43, no. 6, pp.509-521, Nov. 2013.
[24]	D. Cook and M. Schmitter-Edgecombe, "Assessing the quality of activities in a smart environment," Methods Inf. Med., vol. 48, no. 5, pp. 480–485, May 2009.
[25]	H. Chen, S. Zhao, C. Shang, G. Chen, and C. Y. Chang (in press), "Activity recognition approach based on spatial-temporal constraints for aged-care in smart home, " International Journal of Ad Hoc and Ubiquitous Computing, vol. 33, no. 3, pp. 168-183, Apr. 2020.
[26]	C. Shang, C. -Y. Chang, G. Chen, S. Zhao and H. Chen, "BIA: Behavior Identification Algorithm Using Unsupervised Learning Based on Sensor Data for Home Elderly," IEEE Journal of Biomedical and Health Informatics, vol. 24, no. 6, pp. 1589-1600, June 2020.
[27]	L. Yao, et al., "Compressive representation for device-free activity recognition with passive rfid signal strength," IEEE Trans. Mobile Comput., vol. 17, no. 2, pp. 293–306, Feb. 2018.
[28]	Y. Wang, K. Wu, and L. M. Ni, "Wifall: Device-free fall detection by wireless networks," IEEE Trans. Mobile Comput., vol. 16, no. 2, pp. 581-594, Feb. 2017.
[29]	S. Aminikhanghahi, T. Wang, and D. J. Cook, "Real-time change point detection with application to smart home time series data," IEEE Trans. Knowledge Data Eng., vol. 31, no. 5, pp. 1010-1023, May 2018. 
[30]	L.Gutiérrez-Madroñal, et al., "Test event generation for a fall-detection IoT system," IEEE Internet Things J., vol. 6, no. 4, pp. 6642-6651, Aug. 2019.
[31]	H.-H. Hsu and C.-C. Chen, "Rfid-based human behavior modeling and anomaly detection for elderly care," Mobile Inform. Syst., vol. 6, no. 4, pp. 341-354, Dec. 2010.
[32]	M. Nov´ak, F. Jakab, and L. Lain, "Anomaly detection in user daily patterns in smart-home environment," J. Sel. Areas Health Inform., vol. 3, no. 6, pp. 1-11, June 2013.
[33]	N. Dey, A. S. Ashour, F. Shi, S. J.s Fong, and R. S. Sherratt, "Developing residential wireless sensor networks for ECG healthcare monitoring," IEEE Trans. Consumer Electron., vol. 63, no. 4, pp. 442-449, Nov. 2017. 
[34]	L. Fanucci et al., "Sensing devices and sensor signal processing for remote monitoring of vital signs in CHF patients," IEEE Trans. Instrum. Measu., vol. 62, no. 3, pp. 553-569, Mar. 2013 ,
[35]	A. M. Nia et al., "Energy-efficient long-term continuous personal health monitoring," IEEE Trans. Multi-Scale Comput. Syst., vol. 1, no. 2, pp. 85-98, Apr. 2015.
[36]	M. Yu, A. Rhuma, S. Naqvi, L. Wang, and J. Chambers, "A posture recognition-based fall detection system for monitoring an elderly person in a smart home environment," IEEE Trans. Inf. Technol. Biomed., vol. 16, no. 6, pp. 1274-1286, Dec. 2012.
[37]	J. Wang et al., "An enhanced fall detection system for elderly person monitoring using consumer home networks," IEEE Trans. Consumer Electron., vol. 60, no. 1, pp. 23-29, Feb. 2014.
[38]	B. R. Greene, S. J. Redmond, and B. Caulfield, "Fall risk assessment through automatic combination of clinical fall risk factors and body-worn sensor data," IEEE J. Biomed. Health Inform., vol. 21, no. 3, pp. 725-731, Mar. 2017.
[39]	Q. Guan, C. Li, X. Guo, and B. Shen, "Infrared signal based ederly fall detection for in-home monitoring," 9th Int. Conf. Intell. Human-Machine Syst. Cybern., pp. 373-376, 2017.
[40]	D. Elbert et al., "An approach for detecting deviations in daily routine for long-term behavior analysis", 5th Int. Conf. Pervasive Comput. Technol. Healthcare, pp. 426-433, 2011.
[41]	A. R. M. Forkan et al., "A context-aware approach for long-term behavioural change detection and abnormality prediction in ambient assisted living," Pattern Recognit., vol. 48, no. 3, pp. 628-641, Mar. 2015.
[42]	C. J. Shang, et al., "Implicit irregularity detection using unsupervised learning on daily behaviors," IEEE J. Biomed. Health Inform, vol. 7. no. 1, Jan. 2020 .
[43]	C. J. Shang, C. Y. Chang, J. J. Liu, S. H. Zhao, D. S. Roy, "FIID: Feature-Based Implicit Irregularity Detection Using Unsupervised Learning from IoT Data for Homecare of Elderly, " IEEE Internet of Things Journal, vol. 7, no. 11, pp. 10884-10896, Nov. 2020.