Jürgen Perl

Tactical pattern recognition in soccer games by means of special self-organizing maps.

Increasing amounts of data are collected in sports due to technological progress. From a typical soccer game, for instance, the positions of the 22 players and the ball can be recorded 25 times per second, resulting in approximately 135.000 datasets. Without computational assistance it is almost impossible to extract relevant information from the complete data. This contribution introduces a hierarchical architecture of artificial neural networks to find tactical patterns in those positional data. The results from the classification using the hierarchical setup were compared to the results gained by an expert manually classifying the different categories. Short and long game initiations can be detected with relative high accuracy leading to the conclusion that the hierarchical architecture is capable of recognizing different tactical patterns and variations in these patterns. Remaining problems are discussed and ideas concerning further improvements of classification are indicated.

Analysis and simulation of creativity learning by means of artificial neural networks

The paper presents a new neural network approach for analysis and simulation of creative behavior. The used concept of Dynamically Controlled Neural Gas (DyCoNG) entails a combination of Dynamically Controlled Network [Perl, J. (2004a). A neural network approach to movement pattern analysis. Human Movement Science,23, 605-620] and Growing Neural Gas (Fritzke, 1995) by quality neurons. A quality neuron reflects the rareness of a piece of information and therefore can measure the originality of a recorded activity that was assigned to the neuron during the network training. The DyCoNG approach was validated using data from a longitudinal field-based study. The creative behavior of 42 participants in standardized test situations was tested in a creative training program lasting six months. The results from the DyCoNG-based simulation show that the network is able to separate main process types and reproduce recorded creative learning processes by means of simulation. The results are discussed in connection with practical implications in team sports and with a view to future investigations.

Game creativity analysis using neural networks

Abstract Experts in ball games are characterized by extraordinary creative behaviour. This article outlines a framework for analysing types of individual development of creative performance based on neural networks. Therefore, two kinds of sport-specific training programme for the learning of game creativity in real field contexts were investigated. Two training groups (soccer, n = 20; field hockey, n = 17) but not a control group (n = 18) improved with respect to three measuring points (P < 0.001), although no difference could be established between the two training groups (P = 0.212). By using neural networks it is now possible to distinguish between five types of learning behaviour in the development of performance, the most striking ones being what we call “up-down” and “down-up”. In the field hockey group in particular, an up-down fluctuation process was identified, whereby creative performance increases initially, but at the end is worse than in the middle of the training programme. The reverse down-up fluctuation process was identified mainly in the soccer group. The results are discussed with regard to recent training explanation models, such as the super-compensation theory, with a view to further development of neural network applications.

PerPot: A metamodel for simulation of load performance interaction

A metamodel is introduced, which on one hand can help to understand particular effects and phenomena in the interaction of load and performance in training processes. On the other hand, it can be used as a starting point for refinements to specific adaptation models. Finally, a software tool has been developed that supports different simulation approaches—for example, basic analysis of model parameter influences, diagnosis of the state of real adaptation systems, optimization of given load performance interactions, and planning optimal training schedules.

Analysis of Game Creativity Development by Means of Continuously Learning Neural Networks

Experts in ball games are characterized by extraordinary creative behavior. This article outlines a framework of analyzing creative performance based on neural networks. The aim of this study is to compare the potential of different kinds of training programs with the learning of game creativity in real field contexts. The training groups (soccer group, n=20; field hockey group, n=17) showed significant improvement in comparison to the control group (n=18) with respect to the three measuring points, although no difference could be established between the groups. As regards the development of performance, five types of learning behavior can be distinguished, the most striking ones being what we call “up-down” and “down-up”. In the field hockey group in particular, an up-down fluctuation process was identified, whereby the creative performance increases initially, but at the end is worse than in the middle of the training session. The reverse down-up fluctuation process was identified mainly in the soccer group. The results are discussed with regard to recent training explanation models, such as the super-compensation theory, with a view to future investigation.

Special issue: Network approaches in complex environments

With the enormous improvement of data acquisition during the last 10 years or so, the rather serious problem arises how to transform all those data to useful information. Moreover, without analyzing the data one does not know if it contains any useful information at all. Therefore, new and, if possible, automatic devices are needed which are able to accomplish this transformation from data to useful information. Artificial neural networks have proved to be able to do so. As experience from the last decade shows, such networks can successfully be used in sports in case of process analyses where ‘process’ mainly means motion, game, or training. It has to be pointed out that the development of net-based analysis in sports cannot completely be understood from dealing with only one of those fields of application. A lot of correspondences and connections between those different approaches helped to find appropriate solutions in cases that first looked quite different and eventually turned out to be quite similar. The first approaches we are aware of started in the early 2000s with lower level phase analyses of games. Discussions with biomechanics then showed that, reduced to sequences of data, movements are rather similar to games, resulting in fruitful approaches of net-based motion analysis. One main result from this research was that the reduction from the process as a whole to its phases or components may be extremely helpful for understanding and comparing motions. Currently, net-based phase-analysis is used successfully for high level tactic analyses in games. In the context of this background, the special issue is introduced with some remarks on history and technical aspects of net-based process analysis, followed by a brief characterization of the main working areas and closed by a short introduction to the actual contributions to this special issue.

Soccer analyses by means of artificial neural networks, automatic pass recognition and Voronoi-cells: An approach of measuring tactical success.

Success in a soccer match is usually measured by goals. However, in order to yield goals, successful tactical pre-processing is necessary. If analyzing a match with the focus on “success”, promising tactical activities including vertical passes with control win in the opponent’s penalty area have to be the focus. Whether or not a pass is able to crack the opponent’s defence depends on the tactical formations of both the opponent’s defence and the own offence group.

A Pilot Study on Offensive Success in Soccer Based on Space and Ball Control – Key Performance Indicators and Key to Understand Game Dynamics

Abstract The intention of Key Performance Indicators (KPI) is to map complex system-behaviour to single values for scaling, rating and ranking systems or system components. Very often, however, this mapping only reduces important information about tactical behaviour or playing dynamics without replacing it by useful ones. The presented approach tries to bridge the gap between complex dynamics and numerical indicators in the case of offensive effectiveness in soccer in two steps. First, a model is developed which visualises offensive actions in a process-oriented way by using information units to represent offensive performance – i.e. Key Performance Indicators. Second, this model is organised in relation to time intervals, which enables to measure the effectiveness for a whole half-time as well as for arbitrary intervals of any desired lengths. This contribution is meant as an introduction to a new modelling idea, where examples are calculated as case studies to demonstrate how it works. Therefore, only two games have been exemplarily analysed yet: The first one, which is used to demonstrate the method, is an example for similar quantitative indicators but different dynamic behaviour. The last one is used to demonstrate the results in the case of teams with extreme different strengths.

Evaluation of changes in space control due to passing behavior in elite soccer using Voronoi-cells

A soccer player’s ability to make an “effective” pass in a play situation is considered one of the key skills characterizing successful performance in elite soccer.

Ein Expertensystem mit integrierter Simulationskomponente am Beispiel des Tennis-Simulations-Systems TESSY

Das TESSY-Expertensystem organisiert Spiel- bzw. Leistungsdaten von Tennisspielern in einer Datenbank, wertet sie aus, simuliert Spiele zur interaktiven Nutzung auf dem Bildschirm und simuliert Spiele fur die interne Gewinnung zusatzlicher Daten und Fakten.