This presentation was made during the 2004 Pre-Olympic Congress in
Thessaloniki, Greece, August 6-11th. The full text of the article is
currently under review and will be available in the IJCSS in the near
future.
Natalie Balagué & C. Torrents
INEFC. University of Barcelona, Spain
Non-linear relations and interactions between variables characterise
and explain main phenomena occurring in exercise and sport performance.
Nevertheless, the research tools commonly applied are based almost exclusively
upon a simple and linear understanding.
Current training theory and common methods applied by coaches in practice
are also based on the classical science paradigm, that has dominated
and is still influencing sport science very deeply.
The poor and often confusing explanations or solutions for practitioners
and scientists given by the available research results are not only
limiting the development of sport science but also endangering the well-being
of the athlete [1].
The posed question is: is it possible to handle complexity?
Besides showing the limitations of the classical paradigm the aim of
this presentation is to highlight how tools and concepts from computer
science can face the complex nature of sport performance.
Especially alternative modelling and the biologically motivated paradigms
(fuzzy logic, neural networks -ANN, genetic algorithms) allow to:
- integrate interdisciplinary research adopting
a system view [2], avoiding the increasing analytical tendency in
sport sciences,
- getting qualitative information of dynamic processes,
allowing thinking before computing,
- compressing and handling with imprecise information,
instead of simplifying or avoiding it,
- studying interactions among non linear variables
and non linear processes (instead of transforming them in linear or
simple ones) [3].
- develop systems able to learn and generalise the
learn matter (supervised ANN),
- learn, recognise unknown structures and classify
pattern responses (unsupervised ANN -KFM) [5],
- describe, analyse and evaluate continuous adaptation processes
(dynamic controlled networks) [4].
In conclusion, the application of modelling techniques and mainly the
so-called soft computing are suitable tools for understanding the non-linear
complex dynamic processes involved in performance.
References
[1]. Balagué, N. et al. (2002). Acta Ac. Ol. Estonia, 9, 51-63.
[2]. Lames, M. (2003). IJCSS, Vol2,Ed.1
[3]. Perl, J. & Mester, J. (2001). Leistungssport, 2, 54-62.
[4]. Perl, J. (2001). Sport & Informatik, Köln: Straus.
[5]. Schöllhorn, W. &Bauer, H-U. (1998). Informatik im Sport,
Köln: Straus.
http://www.icsspe.org/portal/bulletin-january2005.htm
From Simple to Complex understanding of perofrmance
by Alternative Modelling