We investigate how to learn functions that rate game situations on a soccer pitch according to their potential to lead to successful attacks. We follow a purely data-driven approach using techniques from deep reinforcement learning to valuate multiplayer positionings based on positional data. Empirically, the predicted scores highly correlate with dangerousness of actual situations and show that rating of player positioning without expert knowledge is possible.
Get full access to this article
View all access options for this article.
References
1.
LinkD, LangS, SeidenschwarzP. Real time quantification of dangerousity in football using spatiotemporal tracking data. PLoS One. 2016; 11:e016876–8.
BialkowskiA, LuceyP, CarrP, et al.Large-scale analysis of soccer matches using spatiotemporal tracking data. In: 2014 IEEE International Conference on Data Mining, Shenzhen, 2014, 725–730. doi:10.1109/ICDM.2014.133
4.
LuceyP, BialkowskiA, MonfortM, et al.Quality vs quantity: Improved shot prediction in soccer using strategic features from spatiotemporal data. In: Proc. 8th Annual MIT Sloan Sports Analytics Conference. Boston, MA, 2014, pp. 1–9.
5.
MnihV, BadiaAP, MirzaM, et al.Asynchronous methods for deep reinforcement learning. In: Proceedings of Machine Learning Research (PMLR), New York, 2016, pp. 1928–1937.
6.
MnihV, KavukcuogluK, SilverD, et al.Human-level control through deep reinforcement learning. Nature. 2015; 518:529–533.
7.
SilverD, HuangA, MaddisonCJ, et al.Mastering the game of go with deep neural networks and tree search. Nature. 2016; 529:484–489.
8.
CervoneD, DAmourA, BornnL, GoldsberryK. A multiresolution stochastic process model for predicting basketball possession outcomes. J Am Stat Assoc. 2016; 111:585–599.
9.
CopeteJL, SuzukiJ, WeiQ, et al.Estimation of players actions in soccer matches based on deep autoencoder. Japanese Society for Artificial Intelligence, Technical Report, 2015, p. 7.
10.
FernandoT, WeiX, FookesC, et al.Discovering methods of scoring in soccer using tracking data. Sidney, Large-Scale Sports Analytics, 2015.
11.
Van HaarenJ, HannossetS, DavisJ. Strategy discovery in professional soccer match data. In: Proceedings of the KDD-16 Workshop on Large-Scale Sports Analytics, New York: ACM, 2016.
12.
WangQ, ZhuH, HuW, et al.Discerning tactical patterns for professional soccer teams: An enhanced topic model with applications. In: Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, New York: ACM, 2015. pp. 2197–2206.
13.
HaaseJ, BrefeldU. Mining positional data stream. In: AppiceA, CeciM, LoglisciC, et al. (Eds.): New Frontiers in Mining Complex Patterns, Cham, Switzerland: Springer, 2015, pp 102–116.
14.
Van HaarenJ, DzyubaV, HannossetS, DavisJ. Automatically discovering offensive patterns in soccer match data. In: International Symposium on Intelligent Data Analysis, Cham, Switzerland: Springer, 2015, pp. 286–297.
15.
LuceyP, OliverD, CarrP, et al.Assessing team strategy using spatiotemporal data. In: Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining, New York: ACM, 2013, pp. 1366–1374.
16.
BrandtM, BrefeldU. Graph-based approaches for analyzing team interaction on the example of soccer. In: Proceedings of the ECML/PKDD Workshop on Machine Learning and Data Mining for Sports Analytics, Cham, Switzerland: Springer, 2015.
17.
ReinR, MemmertD. Big data and tactical analysis in elite soccer: Future challenges and opportunities for sports science. SpringerPlus. 2016; 5:141–0.
18.
BradleyAE. The use of the area under the ROC curve in the evaluation of machine learning algorithms. Pattern Recognit. 1997; 30:1145–1159.
19.
van SeijenH, SuttonRS. True online TD (lambda). In: Proceedings of Machine Learning Research (PMLR), Beijing, China, 2014, pp. 692–700.
20.
GlorotX, BordesA, BengioY. Deep sparse rectifier neural networks. In: Proceedings of Machine Learning Research (PMLR), Sardinia, Italy, 2011.
21.
BertsekasDP. Approximate policy iteration: A survey and some new methods. J Control Theory Appl. 2011; 9:310–335.
22.
KocsisL, SzepesváriC. Bandit based monte-carlo planning. In: European conference on machine learning, Cham, Switzerland: Springer, 2006, pp. 282–293.
TesauroG. Temporal difference learning and TD-gammon. Commun ACM. 1995; 38:58–68.
25.
SuttonRS, BartoAG, et al.Reinforcement learning: An introduction. Cambridge, MA: MIT Press, 1998.
26.
HarbJ, PrecupD. Investigating recurrence and eligibility traces in deep Q-networks. NIPS 2016, Deep Reinforcement Learning Workshop, 2016.
27.
HausknechtM, StoneP. Deep recurrent q-learning for partially observable MDPs. In: 2015 AAAI Fall Symposium Series, Palo Alto, CA: AAAI PRESS, 2015.
28.
FoersterJ, AssaelYM, de FreitasN, WhitesonS. Learning to communicate with deep multi-agent reinforcement learning. In: Advances in Neural Information Processing Systems, Barcelona, Spain: NIPS, 2016, pp. 2137–2145.
29.
HeK, ZhangX, RenS, SunJ. Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778.
30.
KingmaD, BaJ. Adam: A method for stochastic optimization. In: Proceedings of the 3rd International Conference on Learning Representations (ICLR), 2015. arXiv preprint. 2014;arXiv:1412.6980.
