运动科学综述阅读(7): Computational intelligence in sports

本文是另一篇Computational intelligence in sports:Challenges and opportunities within a new research domain。 这篇被之前(1~6)中的作者认为很有价值，所以从这里入手阅读。作者提出了一种Programming Trainer Model, 并讨论了一些CI算法。 下一步打算自己试着提取和分析TCX格式的文件，看一看怎么进行分析。 至少可以先试试画图，哈哈

Computational intelligence in sports

Challenges and opportunities within a new research domain

Abstract

computational intelligence(CI): AI + algorithms inspired by nature

● nature inspired algorithms

○ characteristic: collective intelligence and adaptability to a changing environment

○ efficiency and simplicity

○ employed in problems about social and natural sciences

Aim of this paper:

demonstrate nature-inspired algorithms are useful within the domain of sport, in particular for obtaining safe and effective training plans targeting various aspects of performance

Introduction

CI represents algorithms for solving real-world problems somewhat intelligently as similar problems are solved by natural systems

● intelligence by humans refers to a mental adaptation to new circumstances

● 人类的智能是适应新环境

● artificial intelligence is based on an intelligent algorithms’s capability of adapting to changing environment

● 人工智能是基于这些智能算法的适应环境的能力

CI algorithms:

1. artificial neural network ANN

2. evolutionary algorithms EA

3. swarm intelligence SI

4. artificial immune systems AIS

5. fuzzy systems FS

● commonality is the principles for these algorithms for the operations are borrowed from natural systems

Background information(theory of sports training)

● huge amount of data produced by various devices

● training sessions are planned, analyzed and monitored by coaches

● tasks of effective coach:

○ trust in relation with trainees

○ manage the trainee to become a personality

○ not to hurt trainees either physically or psychologically

● knowledge from broad areas: anatomy, physiology, biomechanics, psychology, sociology, and didactics

● purpose of sports’ training: maximal performance, constraints of selected sports’ disciplines

• physical: motor ability. refers to characteristics, like endurance, speed, force, coordination and flexibility
• technical: important skills for team sports of coordination and flexibility
• tactical: different strategies
• psychological: athlete’s personality, like motivation, temperament, individual’s qualities and attitudes

● external factors

● internal factors

● type according to presence of oxygen for breath(metabolism)

○ aerobic: chemical reaction in the muscles is performed with the presence of oxygen. Emit carbon dioxide and water as by products of breathing

○ anaerobic: arise when the intensity of a workout increases above the so-called aerobic threshold. generate lactic acid as by-product, causes fatigue and is difficult to remove from muscles

Computational intelligence in sport

● sports training is

○ to take care of athlete’s psycho-physical perfomances

○ improve the athlete’s technical-tactical performances

● sports’ training is

● planed by a coach

● adapted on the basis of analyzed data obtained during the training sessions

This overview steps further for proposing an artificial personal trainer resembling real coaches with taking advantage of characteristic CI applications

Trainer Model

programming model of the artificial personal trainer

components

1. identifying long-term objectives: represent training strategies <- input: off-line plan(awareness of external factors, internal facotrs and predisposition on an athlete)

● the plan can consist of various types of training sessions like endurance, speed, force and coordination

2. Identifying short-term objectives: represent training tactics -> output: on-line planning, regarding the factors like weather, nutrition and athlete’s feelings.

某种程度是每天课程的意思

3. Data analysis: preprocessing, processing, visualization

a. preprocessing: XML -> item in database

b. processing: data mining and statistical analysis

c. visualization: to present additional, mostly hidden information about training sessions, like familiarization with their habits during training activities, identification of relations between various types of trainings, influence of sleepwear or nutrition on athletes’ readiness

data processing

datafor analyzing are accumulated by sports trackers(on smartphones) or sport watches.

• worn during sports’ training, to collect data
  • internal factors(heart-rate)
  • external factors(distance(measured by GPS, accuracy up to 3 meters for commercial use), duration and altitude)
• Collected data are saved as XML(TCX) files.
• detailed measures about the time-stamp, the latitude(in degrees), the longitude(in degrees), the altitude(in meters), the distance from the last measured position(in meters), the heart-rate(in beats/minute, bps) and the speed(in meters/second) during each tracking interval (alsotrack-point) which lasts a few seconds
• 也就是说，实际存储的是一点一点的数据。两点确定了一段的数据，这样再拟合出整个运动过程。
• each track-point→ performance over second intervals
• huge amount of data, files are produced, espeecially for multi-sport training sessions(e.g. triathlon)

Data processing

influenced by:

1. statistical methods and data visualization
2. artificial intelligence applying the machine learning as well as computational intelligence
3. database system providing methods for storing a huge amount of data and retrieving them from big data warehouses

two types of applications:

1. predictive: predict one or more variables in the future by using a part of the variables in the database
   • classification, regression
2. identify the patterns for describing data stored in the databases and visualizing the data in a way that can be easily understood by the users
   • clustering, association rule mining and sequential pattern discovering

Data visualization

• humans are visual beings
• graphical techniques to tell more information than alphanumerical data alone
• visualization techniques depends on application type
  • query independent: visualize dataset direcly, i.e. without any assertion
  • query dependent: query sepcified by a user
• a novel technique uusing glyphs can be used for illustrating sequential patterns or time sequences within big datasets

Glyphs are graphical entities which convey one or more data values via attributes like shape, size, color and position

• a geometric projection techniques search for a convenient projection of data in multiple dimensions

Existing applications within the sports domain

applications of computational intelligence in sports, and also some applications from this broad spectra of a research area

categorized by the covered issues

1. managing the sports training
2. sports training and the pervasive computing
3. other applications in sports

Managing the sports training

Table 1 Managing the sports training.

Application	Reference
Analyzing of performances in sports	[40]
Rapid feedback systems	[31]
Adaptive systems in sport	[29]
Modeling of training loads	[51]
Automatic physical effort plan generation	[36]
Sport training modelling	[42]
Recruitment process for sport swimming	[50]
Complex systems in sport	[34]
Automatic evaluation of exercises	[43]
Training optimization	[61]
Sport training support	[35,37]
Method and system of delivering an interactive and dynamic multi-sport training program	[59]
Performance evaluation	[46]

TCX格式的运动文件中有大量的数据，不方便人类处理。

作者提到了

1. [Fister et al.](I. Fister Jr., D. Fister, I. Fister, S. Fong, Data mining in sporting activities created by sports trackers, Comput. Bus. Intell. (ISCBI), 2013 Int. Symp., August 2013, pp. 88–91.)

   dealt with the problem of how to exploit data obtained from sports watches in TCX format

2. [Fister et al.](I. Fister, S. Rauter, X.-S. Yang, K. Ljubic, I. Fister Jr., Planning the sports training sessions with the bat algorithm, Neurocomputing Part B 149 (0) (2015) 993–1002.)

   bat algorithm for planning the number of sports sessions

3. [AI in sports](H. Novatchkov, A. Baca, Artificial intelligence in sports on the example of weight training, J. Sports Sci. Med. 12 (1) (2013) 27.)

   the potential of artificial intellignece techniques in sports using the example of weight training

4. [fuzzy logic for evaluate strength training exercises] H. Novatchkov, A. Baca, Fuzzy logic in sports: a review and an illustrative case study in the field of strength training, Int. J. Comput. Appl. 71 (6) (2013) 8–14.

5. [computing athletic performance] P. Skiba, System and method for computing athletic performance, US Patent 8,533,001.2013.

Sports training and pervasive computing

a bunch of applications for monitoring sporting activities, for predicting proper food, for helping people lose weight and even for envouraging people to practise sport; a number of academic papers have arisen over the past years.

Table 2 Sports training and pervasive computing

Application	Reference
A server-based mobile coaching system	[33]
Wearable system for fitness training	[38]
A mobile health application for dietary evaluation	[52]
Motivation	[28,62]
Fitness companion demonstrator	[55]
Running performance	[56]
Exercise repetition detection	[48]
Motion rehabilitation training system	[54]
Filtering fitness trail content	[39]
Physical activity support [27	,57]
Music synchronizer	[41]

A comprehensive survey about the influence of ubiquitous computing in sports is presented in [comprehensive survey]([30] A. Baca, P. Dabnichki, M. Heller, P. Kornfeind, Ubiquitous computing in sports: a review and analysis, J. Sports Sci. 27 (12) (2009) 1335–1346.)

Other applications in sports

too special to be categorized into the mentions types

Table 3
Other research and applications in sports domain.

Application	Reference
Stability analysis of motion patterns in biathlon shooting	[32]
Jump detection	[49]
Nonlinear fuzzy control of human heart rate	[47]
Ambient intelligence systems for personalized sport training	[60]
Peloton phase oscillations	[58]
Problem of nutrition during sport competitions	[9]
Avoiding over-training	[12]

Future ideas and opportunities for research

1. mobile devices have been supplemented with sensors that are capable of identifying indicators like: lactate, oxygen consumption and respiratory exchange ratio(RER). Power-meters are developed especially for cycle training.

   硬件很丰富

2. 这个领域研究的基本点是data mining. 但是传统的数据挖掘方法不能应对该领域复杂的问题。 所以，CI data mining方法弥补了传统方法的不足。 这些CI算法的终极目标是实现一个人工智能的教练，在所有领域替代传统的人类教练，并且价格更便宜。 这对于业余运动员，尤其是能够和职业运动员一起比赛的那些项目的运动员（例如铁人三项），更加重要。

   An artificial trainer would provide all the services of a human coach and would be cheaper. This is especially important for amateur athletes whose counts have increased due to emergence of multi-disciplinary sports like triathlon, where amateur athletes compete with the professionals in the same event at the same time.
   为什么更重要？

3. 避不开的领域

   1. clustering. simplify large, complex TCX files
   2. classification.
   3. sequential pattern mining. detect trends during training period
   4. association rule mining. identify the athletes’ habits in training

4. 除了教练，还可以应用的领域

   1. Optimization of sports’ equipment like bike frames, wetsuits and helmets: EAs and SI-based algorithms are suitable for the optimization of sport equipment. For instance, road bikers need specific bikes that can be adjusted using CI algorithms.
   2. Optimization of different player positions on the field in team sport: Here, a task of CI algorithm is to determine which position in the field is the more suitable for a specific player.
   3. Detection of doping in sport: Doping can easily be detected by an artificial personal trainer from historical files about sports activities using data mining algorithms.
   4. Training course generation: Choosing a proper course for training is still a very hard task for athletes. Why not develop a CI algorithm for generating a full training course by specifying some requirements and constraints?
   5. Predictions of overall times in different competitions: Based on results from previous training sessions and using the personal history of an athlete, the CI algorithm can predict an overall time before competition takes place.
   6. Detection of an athlete’s crisis during endurance competitions: Almost every athlete encounters a crisis during an endurance race. Using the sequential pattern mining, the artificial personal trainer can find conditions where an athlete usually falls into a crisis, how long it is, and how to minimize this phenomenon by an athlete.
   7. Virtual simulations may also predict the behavior of our opponents: In other words, currently it is easy to fetch historical data of athletes during the years from the Internet that also enables us to analyze the data of our opponents. From these data, the CI algorithm can mine characteristics of our opponents, and searches for their potential weaknesses that can be used against them.
   8. Avoiding pains and over-training: Using the data mining methods, this is an easy task that may prevent a trainee from
      overdoing sports’ training.

Conclusion

1. artificial personal trainer can
   1. generate a plan for proper training sessions according to an agreed training strategy
   2. adapt the training plan according to current external and internal factors
   3. measure indicators of an athlete’s readiness during the training sessions from different sources like heart-rate, lactate, RER and power-meter
   4. analyze the huge data produced by mobile tracking devices during training sessions
2. convert some usability of the proposed artificial personal trainer to team sports. And Swarm Intelligence algorithms seem to be more suitable for collective sports, where the interplay among many cooperative might lead to more sophisticated behaviors and unexpected outcomes might arguably be obtained.

Appendix

A. Sport activities dataset

lack of dataset of sport activities

[one comprehensive dataset](S. Rauter, I. Fister Jr., I. Fister, A Collection of Sport Activity Files for Data Analysis and Data Mining, Technical report 0201, University of Ljubljana and
University of Maribor, 2015.) released and covered nine persons, which covers all of their training and its speciality is that their entire seasons are visible.

B. Call for sport activities data for research

athletes to donate, security from comparisons or espionage

Reference

References
[1] A.R.R. Alazmi, A.A.R. Alazmi, Data mining and visualization of large databases, Int. J. Comput. Sci. Secur. 6 (5) (2012) 295–314.
[2] C. Blum, D. Merkle, Swarm Intelligence, Natural Computing Series, Springer Verlag, Berlin, 2008.
[3] X. Chen, Y.-S. Ong, M.-H. Lim, K.C. Tan, A multi-facet survey on memetic computation, IEEE Trans. Evol. Comput. 15 (5) (2011) 591–607.
[4] D.H.S. Chung, P.A. Legg, M.L. Parry, R. Bown, I.W. Griffiths, R.S. Laramee, M. Chen, Glyph sorting: Interactive visualization for multi-dimensional data, Inform.
Visual. 14 (2015) 76–90.
[5] S. Das, P.N. Suganthan, Differential evolution: A survey of the state-of-the-art, Evol. Comput. IEEE Trans. 15 (1) (2011) 4–31.
[6] R.C. Eberhart, Computational Intelligence: Concepts to Implementations, Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 2007.
[7] A.E. Eiben, J.E. Smith, Introduction to Evolutionary Computing, Natural Computing Series, Springer Verlag, Berlin, 2003.
[8] R.A. Elmasri, S.B. Navathe, Fundamentals of Database Systems, third ed., Addison-Wesley Longman Publishing Co., Inc., Boston, MA, USA, 1999.
[9] I. Fister, D. Fister, K. Ljubic, Y. Zhuang, S. Fong, Towards automatic food prediction during endurance sport competitions, in: Proceedings of the 2014
International Conference on Soft Computing & Machine Intelligence, IEEE Computer Society Press, 2014, pp. 6–10.
[10] I. Fister Jr., D. Fister, I. Fister, S. Fong, Data mining in sporting activities created by sports trackers, Comput. Bus. Intell. (ISCBI), 2013 Int. Symp., August 2013,
pp. 88–91.
[11] I. Fister, S. Rauter, X.-S. Yang, K. Ljubic, I. Fister Jr., Planning the sports training sessions with the bat algorithm, Neurocomputing Part B 149 (0) (2015) ˇ
993–1002.
[12] I. Fister Jr., G. Hrovat, I. Fister, S. Rauter, Am I Overtraining? A Novel Data Mining Approach for Avoiding Overtraining, in: Proceedings of the 2014 1st Student
Computer Science Research Conference, Primorska Press, 2014, pp. 45–52.
[13] S. Forrest, A.S. Perelson, L. Allen, R. Cherukuri, Self-nonself Discrimination in a Computer, in: Proceedings of the 1994 IEEE Symposium on Research in
Security and Privacy, IEEE Computer Society Press, 1994, pp. 202–212.
[14] J. Gaudette, The Ultimate Marathon ebook. December 2014,
186 I. Fister Jr. et al. / Applied Mathematics and Computation 262 (2015) 178–186
[15] F. Gorunescu, Intelligent systems reference library, Data Mining: Concepts, Models and Techniques, Springer, Berlin, Hiedelberg, 2011.
[16] R. Kovahi, F. Provost, Glossary of terms, Mach. Learn. 1 (30) (1998) 271–274.
[17] H. Müller, R. Reihs, K. Zatloukal, A. Holzinger, Analysis of biomedical data with multilevel glyphs, BMC Bioinform. 15 (6) (2014) S5 Suppl.
[18] J. Piaget, The Psychology of Intelligence. Routledge Classics, 2001.
[19] S. Russell, P. Norvig, Artificial Intelligence: A Modern Approach, Prentice Hall, New Jersey, 2009.
[20] C.A. Rüst, M.A. Zingg, T. Rosemann, B. Knechtle, Will the age of peak ultra-marathon performance increase with increasing race duration? BMC Sports Sci.
Med. Rehabil. 6 (1) (2014) 36.
[21] M.O. Ward, A taxonomy of glyph placement strategies for multidimensional data visualization, Inf. Vis. 1 (3–4) (2002) 194–210.
[22] D. Zahradnik, P. Korvas, The Introduction into Sports Training. Masaryk University, Brno, 2012.
[23] S. Rauter, I. Fister Jr., I. Fister, A Collection of Sport Activity Files for Data Analysis and Data Mining, Technical report 0201, University of Ljubljana and
University of Maribor, 2015.
[24] B. Alatas, E. Akin, Rough particle swarm optimization and its applications in data mining, Soft Comput. 12 (12) (2008) 1205–1218.
[25] C. Grosan, A. Abraham, M. Chis, Swarm Intelligence in Data Mining, Springer, 2015.
[26] K.E. Heraguemi, N. Kamel, H. Drias, Association rule mining based on bat algorithm, in: Bio-Inspired Computing-Theories and Applications, Springer, 2014,
pp. 182–186.
[27] A. Ahtinen, M. Isomursu, S. Ramiah, J. Blom, Advise, acknowledge, grow and engage: Design principles for a mobile wellness application to support physical
activity, Int. J. Mobile Human Comput. Interact. (IJMHCI) 5 (4) (2013) 20–55.
[28] S.M. Arteaga, M. Kudeki, A. Woodworth, S. Kurniawan, Mobile system to motivate teenagers’ physical activity, in: Proceedings of the 9th International
Conference on Interaction Design and Children, ACM, 2010, pp. 1–10.
[29] A. Baca, Adaptive systems in sports, Social Networks and the Economics of Sports, Springer, 2014, pp. 115–124.
[30] A. Baca, P. Dabnichki, M. Heller, P. Kornfeind, Ubiquitous computing in sports: a review and analysis, J. Sports Sci. 27 (12) (2009) 1335–1346.
[31] A. Baca, P. Kornfeind, Rapid feedback systems for elite sports training, Pervasive Comput. IEEE 5 (4) (2006) 70–76.
[32] A. Baca, P. Kornfeind, Stability analysis of motion patterns in biathlon shooting, Hum. Mov. Sci. 31 (2) (2012) 295–302.
[33] A. Baca, P. Kornfeind, E. Preuschl, S. Bichler, M. Tampier, H. Novatchkov, A server-based mobile coaching system, Sensors 10 (12) (2010) 10640–10662.
[34] N. Balague, C. Torrents, R. Hristovski, K. Davids, D. Araújo, Overview of complex systems in sport, J. Syst. Sci. Complex. 26 (1) (2013) 4–13.
[35] K. Brzostowski, J. Drapała, G. Dziedzic, J. Swiatek, Algorithm to plan athletes prolonged training based on model of physiological response, in: Intelligent ´
Information and Database Systems, Springer, 2015, pp. 273–283.
[36] K. Brzostowski, J. Drapała, A. Grzech, P. Swiatek, Adaptive decision support system for automatic physical effort plan generationdata-driven approach, ´
Cybernet. Syst. 44 (2–3) (2013) 204–221.
[37] K. Brzostowski, J. Drapała, J. Swiatek, Application of nonlinear state estimation methods for sport training support, in: Intelligent Information and Database Systems, Springer, 2014, pp. 513–521.
[38] F. Buttussi, L. Chittaro, Mopet: A context-aware and user-adaptive wearable system for fitness training, Artif. Intell. Med. 42 (2) (2008) 153–163.
[39] F. Buttussi, L. Chittaro, D. Nadalutti, Filtering fitness trail content generated by mobile users, in: User Modeling, Adaptation, and Personalization, Springer, 2009, pp. 441–446.
[40] P. Jürgen, B. Arnold, Application of neural networks to analyze performance in sports, in: Proceedings of the 8th Annual Congress of the European College of Sport Science, Salzburg: ECSS, vol. 342, 2003.
[41] T. Kitahara, S. Hokari, T. Nagayasu, Music synchronizer with runners pace for supporting steady pace jogging, in: HCI International 2014-Posters Extended Abstracts, Springer, 2014, pp. 343–348.
[42] E. Me, O. Unold, Machine learning approach to model sport training, Comput. Hum. Behav. 27 (5) (2011) 1499–1506.
[43] H. Novatchkov, A. Baca, Machine learning methods for the automatic evaluation of exercises on sensor-equipped weight training machines, Proc. Eng. 34 2012) 562–567.
[44] H. Novatchkov, A. Baca, Artificial intelligence in sports on the example of weight training, J. Sports Sci. Med. 12 (1) (2013) 27.
[45] H. Novatchkov, A. Baca, Fuzzy logic in sports: a review and an illustrative case study in the field of strength training, Int. J. Comput. Appl. 71 (6) (2013) 8–14.
[46] G. Owusu, Ai and computer-based methods in performance evaluation of sporting feats: an overview, Artif. Intell. Rev. 27 (1) (2007) 57–70.
[47] A. Patrascu, M. Patrascu, I. Hantiu, Nonlinear fuzzy control of human heart rate during aerobic endurance training with respect to significant model
variations, in: System Theory, Control and Computing (ICSTCC), 2014 18th International Conference, October 2014, pp. 311–316.
[48] I. Pernek, K.A. Hummel, P. Kokol, Exercise repetition detection for resistance training based on smartphones, Personal Ubiquitous Comput. 17 (4) (2013) 771–782.
[49] C.L. Roberts-Thomson, A.M. Lokshin, V.A. Kuzkin, Jump detection using fuzzy logic, in: Computational Intelligence for Engineering Solutions (CIES), 2014 IEEE Symposium on, IEEE, 2014, pp. 125–131.
[50] R. Roczniok, I. Rygula, A. Kwasniewska, The use of Kohonen’s neural networks in the recruitment process for sport swimming, J. Hum. Kinet. 17 (2007) 75.
[51] I. Rygula, Artificial neural networks as a tool of modeling of training loads, in: Engineering in Medicine and Biology Society, 2005. IEEE-EMBS 2005. 27th Annual International Conference of the, IEEE, 2006, pp. 2985–2988.
[52] B.M. Silva, I.M. Lopes, J.J. Rodrigues, P. Ray, Sapofitness: A mobile health application for dietary evaluation, in: e-Health Networking Applications and Services Healthcom), 2011, 13th IEEE International Conference on, IEEE, 2011, pp. 375–380.
[53] P. Skiba, System and method for computing athletic performance, US Patent 8,533,001.2013.
[54] G. Spina, G. Huang, A. Vaes, M. Spruit, O. Amft, Copdtrainer: a smartphone-based motion rehabilitation training system with real-time acoustic feedback, in: Proceedings of the 2013 ACM International Joint Conference on Pervasive and Ubiquitous Computing, ACM, 2013, pp. 597–606.
[55] O. Ståhl, B. Gambäck, M. Turunen, J. Hakulinen, A mobile health and fitness companion demonstrator, in: Proceedings of the 12th Conference of the European
Chapter of the Association for Computational Linguistics: Demonstrations Session, Assoc. Computat. Linguist., 2009, pp. 65–68.
[56] C. Strohrmann, H. Harms, G. Troster, What do sensors know about your running performance? in: Wearable Computers (ISWC), 2011, 15th Annual International Symposium on, IEEE, 2011, pp. 101–104.
[57] P. Swiatek, P. Klukowski, K. Brzostowski, J. Drapała,Wearable smart system for physical activity support, in: Advanced Techniques for Knowledge Engi ´ neering and Innovative Applications, Springer, 2013, pp. 291–304.
[58] H. Trenchard, Peloton phase oscillations, Chaos, Solitons & Fract. 56 (2013) 194–201.
[59] R.J. Urbanowski, S.E. Leckie, L.S. Pasin, S.S.Q. Whitfield, Method and system of delivering an interactive and dynamic multi-sport training program, US Patent App. 13/411,437, 2012.
[60] J. Vales-Alonso, P. López-Matencio, F. J. Gonzalez-Castaño, H. Navarro-Hellín, P.J. Baños-Guirao, F.J. Pérez-Martínez, R.P. Martínez-Álvarez, D. GonzálezJiménez, F. Gil-Castiñeira, R. Duro-Fernández, Ambient intelligence systems for personalized sport training, Sensors 10 (3) (2010) 2359–2385.
[61] M.E. van der Zande, J.Van Der Loo, S. Kalisvaart, System for training optimisation, US Patent 8,348,809, 2013.
[62] G. Wijnalda, S. Pauws, F. Vignoli, H. Stuckenschmidt, A personalized music system for motivation in sport performance, IEEE Pervasive Comput. 4 (3) (2005) 26–32.