Responsive image
博碩士論文 etd-0113114-102208 詳細資訊
Title page for etd-0113114-102208
論文名稱
Title
基於智慧型手機行為辨識進行社群活動之推薦
A Smartphone-based Activity Recognition Framework for Social Event Recommendation
系所名稱
Department
畢業學年期
Year, semester
語文別
Language
學位類別
Degree
頁數
Number of pages
66
研究生
Author
指導教授
Advisor
召集委員
Convenor
口試委員
Advisory Committee
口試日期
Date of Exam
2014-01-28
繳交日期
Date of Submission
2014-02-13
關鍵字
Keywords
加速度計、機器學習、事件推薦、智慧型手機、活動辨識
event recommendation, smartphone, activity recognition, accelerometers, machine learning
統計
Statistics
本論文已被瀏覽 5885 次,被下載 742
The thesis/dissertation has been browsed 5885 times, has been downloaded 742 times.
中文摘要
隨著智慧型手機硬體的顯著進步,藉由手機內建的感測器來達到人類活動的辨識已經不再是那麼遙不可及。活動辨識對於醫療保健,健康管理和預防是一項重要的資訊的來源。那些具有創意的活動辨識手機程式,即所謂有能力去判斷目前個人在進行怎樣活動的軟體,給予其用者可以查看他們家人目前的健康及安全情況。

本篇論文旨在達成兩項目標:複雜活動的辨識以及可以促進社群互動的手機程式。為了完成這兩項目標,有些步驟是需要的。首先,對於簡單活動特徵向量的分析。簡單活動的資料像是坐姿、走路、騎機車及搭電梯等訊息被搜集起來並轉換成特徵向量供進一步使用。結果指出,藉由所選的特徵選擇方法,辨識率得到了改善,計算成本也被降到最低。

複雜活動辨識像是搭乘大眾運輸工具、出門購物、放鬆及用餐仍然具有相當的挑戰性。這些複雜活動通常由好幾個簡單活動所組成,而且隨著個體的不同,其變化性也對的大。在這項研究中,共有17個小時的數據記錄。另外,為了降低數據集的數量和基於所創建的模型來表示複雜活動,兩個高階特徵選擇方法,BoF和LDA,被採用。結果顯示,活動標籤結合BoF可以提供比LDA更好的準確率。

最後本篇論文也給予其它研究者有關感應器探勘相關應用程式的考量,對於未來的開發將會有一定的助益。
Abstract
With significant advances made in smartphone hardware, recognizing human activities from smartphone sensors has become possible. Activity recognition is a valuable source of information for healthcare, fitness tracking and prevention. Those innovative activity-recognition-based applications which have the ability to identify the activity a person is performing allow users to monitor the health and safety of their family members.

This thesis aims to achieve two major goals: the recognition of complex activities and an Android-based application for physical social interaction. In order to complete the two goals, some steps are required. Firstly, an analysis of feature selection methods for micro-activity recognition. A data set of micro-activities such as sitting, walking, riding a scooter, driving, and riding an elevator were collected and transformed into feature vectors for further use. The results indicate that with the help of chosen feature selection methods, recognition accuracy was improved and computational cost was also minimized.

Recognition of complex activity such as using public transportation, going out for shopping, relaxing, and having a meal is still very challenging. These activities are composed of several micro activities and vary strongly across individuals. In this study, a total 17 hours of data were recorded. Besides, in order to reduce the volume of dataset and represent complex activities based on a created model, two high-order feature selection methods, BoF and LDA, are considered. The result indicates that BopF combined with recognized activity labels can provide better accuracy than LDA.

In the end, this thesis also gives some thoughts to the design of the sensor mining application which will be useful for others to build their own application.
目次 Table of Contents
摘要 .......................................................................................................... i
Abstract .....................................................................................................ii
Table of Contents ........................................................................................iii
List of Figures .............................................................................................v
List of Tables ..............................................................................................vi

Chapter 1: Introduction .................................................................................1
1.1 Motivation ..............................................................................................1
1.2 Current Problems in Activity Recognition .................................................2
1.3 Objectives .............................................................................................3
1.4 Requirements ........................................................................................4
1.5 Thesis Organization ..............................................................................5
1.6 Chapter Overview ...................................................................................5

Chapter 2: Background ................................................................................7
2.1 Mobile Sensors .....................................................................................7
2.2 Current AR Applications ........................................................................9
2.3 Activity Labeling Techniques .................................................................12
2.4 Data Segmentation Approaches ............................................................12
2.5 Activity Recognition Approaches ..........................................................15

Chapter 3: Proposed Recognition Framework ...............................................21
3.1 Activity Recognition Process .................................................................21
3.2 Sensors Used in This Work ..................................................................21
3.3 Extraction of Relevant Features .............................................................23
3.4 Feature Selection .................................................................................25
3.5 Recognition Framework ........................................................................26

Chapter 4: Micro Activity Recognition ..........................................................28
4.1 Chosen Set of Micro Activities .............................................................. 28
4.2 Data Collection for Micro Activity ..........................................................28
4.2.1 Hardware Platform ............................................................................29
4.2.2 Data Collection ................................................................................29
4.3 Feature Extraction and Analysis ...........................................................31
4.4 Evaluation on Classification Algorithms ..................................................33
4.5 Feature Selection Methods ..................................................................35
4.6 Evaluation on Feature Selection Methods ..............................................36
Chapter 5: Complex Activity Recognition .....................................................39
5.1 Chosen Set of Complex Activities ..........................................................39
5.2 Data Collection for Complex Activity ......................................................40
5.3 High-Order Feature Extraction and Analysis.............................................42
5.4 Evaluation on the Classification Framework .............................................43

Chapter 6: System Design and Implementation .............................................46
6.1 System Design and Architecture ............................................................46
6.2 Cross-Language Communication Mechanism .........................................48
6.3 Data Storage ........................................................................................49

Chapter 7: Conclusions and Future Work .....................................................51
7.1 Conclusions .........................................................................................51
7.2 Future Work ........................................................................................52
參考文獻 References
[1] Fitbit. http://www.fitbit.com
[2] Nike+ Fuelband. http://www.nike.com/fuelband
[3] S. Consolvo et al., “Activity Sensing in the Wild: A Field Trial of Ubifit Garden,” Proc.
26th Annual ACM SIGCHI Conf. Human Factors Comp. Sys., 2008, pp.1797–1806.
[4] A. Cheung, A. Thiagarajan, and S. Madden. “Automatically generating interesting
events with lifejoin”. In proc. Sensys, pages 411–412, 2011.
[5] CHOUJAA, D. AND DULAY, N. Activity recognition using mobile phones:
Achievements, challenges and recommendations. In Proceedings of the Workshop
on How To Do Good Research in Activity Recognition: Experimental Methodology,
Performance Evaluation and Reproducibility in conjunction with UBICOMP, 2010.
[6] A.J. Bernheim Brush, John Krumm, James Scott, and T. Scott Saponas.
Recognizing Activities from Mobile Sensor Data: Challenges and Opportunities.
Microsoft Research, 2011.
[7] J.R. Kwapisz, G. M. Weiss, and S.A. Moore. “Activity recognition using cell phone
  accelerometers,” Proceedings of the Fourth International Workshop on Knowledge
  Discovery from Sensor Data, pp. 10-18, 2010.
[8] A. Rai, Z. Yan, D. Chakraborty, T. Wijaya, and K. Aberer. Mining complex activities
in the wild via a single smartphone accelerometer. In Proceedings of the Sixth
International Workshop on Knowledge Discovery from Sensor Data, pages
  43-51.ACM, 2012.
[9] Yang, J. 2009. Toward physical activity diary: Motion recognition using simple
acceleration features with mobile phones, In First International Workshop on
Interactive Multimedia for Consumer Electronics at ACM Multimedia.
[10] Oreskovic NM, Blossom J, Field AE, Chiang SR, Winickoff JP, Kleinman RE.
Combining global positioning system and accelerometer data to determine the
locations of physical activity in children. Geospat Health. 2012.
[11] L. Liao, D. J. Patterson, D. Fox, and H. Kautz, “Learning and inferring transportation
routines,” Artif. Intell., vol. 171, no. 5–6, pp. 311–331,2007.
[12] S. Kratz, M. Rohs, G. Essl. Combining Acceleration and Gyroscope Data for
Motion Gesture Recognition using Classifiers with Dimensionality Constraints.
IUI'13, 2013.
[13] Q. Li, J.A. Stankovic, M. Hanson, A. Barth, J. Lach. Accurate, fast fall detection
using gyroscopes and accelerometer-derived posture information. In Proceedings
of the 6th International Workshop on Wearable and Implantable Body Sensor
Networks (BSN’09), 2009.
[14] Qin, C., Bao, X., Choudhury, R. R., and Nelakuditi, S.TagSense: A smartphone
based approach to automatic image tagging. In ACM MobiSys , 2011.
[15] H. Lu et al., “Sound-Sense: Scalable Sound Sensing for People-Centric
Applications on Mobile Phones,” Proc. 7th ACM MobiSys, 2009, pp. 165–78.
[16] A. Nishimura, I. Siio, “conteXinger : A Context-aware Song Generator”, Proc.
13th ACM, UbiComp, 2013, pp. 87-90.
[17] M. Fahim, I. Fatima, S. Lee, Y.K. Lee, “Daily Life Activity Tracking Application
for Smart Homes using Android Smartphone”. In IEEE ICACT, 2012.
[18] B. Das, A. M. Seelye, B. L. Thomas, D. J. Cook, L. B. Holder, M. S. Edgecombe,
“Using Smart Phones for Context-Aware Prompting in Smart Environments”. In
EEE CCNC, 2012.
[19] A. Anjum and M. U. Ilyas, “Activity recognition using smartphone sensors,”
Consumer Commun. and Networking Conference (CCNC), 2013 IEEE, pp. 914-919,
Jan. 2013.
[20] C. Qin, X. Bao, R. R. Choudhury and S. Nelakuditi, “TagSense: A smartphone based
approach to automatic image tagging”. In ACM MobiSys (2011).
[21] E. Miluzzo et al., “Sensing meets Mobile Social Networks: The Design,
Implementation, and Evaluation of the CenceMe Application”, Proc. 6th ACM SenSys,
2008, pp. 337–50.
[22] T. Hu`ynh, M. Fritz, B. Schiele, “Discovery of Activity Patterns using Topic Models”,
Proceedings of the Tenth International Conference on Ubiquitous Computing; 2008.
pp. 10–19.
[23] M. Zhang and A. A. Sawchuk, “Motion primitive-based human activity recognition
using a bag-of-features approach,” in ACM SIGHIT International Health Informatics
Symposium (IHI), (Miami, Florida, USA), pp. 1–10, January 2012.
[24] T. Hu`ynh, M. Fritz, B. Schiele, “Discovery of Activity Patterns using Topic Models”,
Proceedings of the Tenth International Conference on Ubiquitous Computing; 2008.
pp. 10–19.
[25] D. J. Cook, N. C. Krishnan, P. Rashidi, “Activity Discovery and Activity Recognition:
New Partnership”, IEEE Transactions on Cybernetics, June, 2013.
[26] L. Breiman, E. Schapire, “Random Forests”, In Machine Learning, pages 5-32 June,
2001.
[27] P. Casale, O. Pujol, P. Radeva, ”Human activity recognition from accelerometer
data using a wearable device” , Proc. 5th ACM, IbPRIA, 2011, pp. 289-296.
[28] S. Dernbach, B. Das, N. C. Krishnan, B. L. Thomas, D. J. Cook, ”Simple and
Complex Activity Recognition Through Smart Phones” , Proc. 8th IEEE, IE, 2012,
pp. 214-221.
[29] D. L. Vail, M. M. Veloso, J. D. Lafferty, ”Conditional random fields for activity
recognition”, Proc. 6th ACM, AAMAS, 2007.
[30] G. M. Weiss and J. W. Lockhart. The Impact of Personalization on Smartphone-
Based Activity Recognition. In AAAI Workshop on Activity Context Representation:
Techniques and Languages, 2012.
[31] G. Dong and J. Li. “Efficient Mining of Emerging Patterns: Discovering Trends and
Differences”, Proc. 5th ACM, SIGKDD, 1999, pp. 43-52.
[32] T. Gu, L. Wang, Z. Wu, X. Tao and J. Lu. “A Pattern Mining Approach to Sensor-
Based Human Activity Recognition”, IEEE, Transactions, 2011, pp.1359-1372.
[33] Z. Yan, D. Chakraborty, A. Misra, H. Jeung and K. Aberer. “Semantic Activity
Classification Using Locomotive Signatures from Mobile Phones”, EPFL-REPORT,
2012.
[34] Z. HE, L. Jin, “Activity Recognition from acceleration data Based on Discrete Consine
Transform and SVM”, Proc. IEEE, Man and Cybernetics, 2009, pp. 5041–5044.
[35] Mallet. http://mallet.cs.umass.edu/
[36] Lockhart, J.W. and Weiss, G. 2011. Design considerations for the WISDM smart
phone-based sensor mining architecture. In Proceedings of the Fifth International
Workshop on Knowledge Discovery from Sensor Data, San Diego, CA.
電子全文 Fulltext
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available


紙本論文 Printed copies
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。
開放時間 available 已公開 available

QR Code