x
Preface
Following computational technologies are currently considered to be crucial for
ecosystems analysis, synthesis and forecasting:
- High performance computing to provide high-speed data access and processing
and large internal storage (RAM), and to facilitate high speed simulations;
- Internet and www to facilitate interactive and onIine simulation as weIl as
software and model sharing;
- Cellular automata to facilitate spatio-temporal and individual-based simulation;
- Fuzzy logic to represent and process uncertain data;
- Artificial neural networks to facilitate multivariate nonlinear regression,
ordination and c1ustering, multivariate time series analysis, image analysis at
micro and macro scale;
- Genetic and evolutionary algorithms for the discovery and evolving of
multivariate nonIinear ruIes, functions, differential equations and artificial neural
networks; - Hybrid and AI models by the embodiment of evolutionary algorithms
in process-based differential equations, the embodiment of fuzzy logic in artificial
neural networks or knowledge processing;
- Adaptive agents to facilitate adaptive simulation and prediction of ecosystem
composition and evolution.
The present book is an outcome of the International Conference on
Applications of Machine Learning to Ecological Modelling, 27 November to 1
December 2000, Adelaide, Australia, which conc1uded with the foundation of the
International
Society
for
Ecological
Informatics
(ISEI)
(http://www.waite.adelaide.edu.auJISEI/). The chapters of the present book are
based on selected papers of the conference, which are exemplary for current
research trends in ecological informatics.
Chapters 1 to 5 address principles and ecological application of fuzzy logic,
artificial neural networks, genetic algorithms, evolutionary computation and
adaptive agents. Salski summarizes concepts of fuzzy logic and discusses
applications for knowledge-based modeling, c1ustering and kriging related to
ecotoxicological, geological and population dynamics data. Giraudel and Lek
discuss the design and appIication of unsupervised artificial neural networks for
the c1assification and visualization of multivariate ecological data. They
demonstrate the potential of Kohonen-type algorithms by c1ustering data of forest
communities in Wisconsin (USA). Morrall discusses origins and nature of genetic
algorithms, and their suitability to induce numericalor rule-based models for
ecological applications. Whigham and Fogel provide a scope of evolutionary
algorithms and their potential for evolving rules, algebraic and differential
equations relevant to ecology. They also address developments on individual and
cooperative behaviour, prey-predator algorithms and hierarchical ecosystems
based on evolutionary algorithms. Recknagel reflects on HoIland's adaptive agents
concept and its potential to more realistically simulate emergent ecosystem
structures and behaviours. He distinguishes between individual-based and state
variable-based agents, and emphasizes on the embodiment of evolutionary
computation in state-variable based agents.
Chapters 6 to 9 provide case studies for the prediction and elucidation of stream
ecosystems by means of machine learning techniques. Goethals, Dedecker,
Gabriels and de Pauw demonstrate applications of c1assification trees and artificial
Preface
Following computational technologies are currently considered to be crucial for
ecosystems analysis, synthesis and forecasting:
- High performance computing to provide high-speed data access and processing
and large internal storage (RAM), and to facilitate high speed simulations;
- Internet and www to facilitate interactive and onIine simulation as weIl as
software and model sharing;
- Cellular automata to facilitate spatio-temporal and individual-based simulation;
- Fuzzy logic to represent and process uncertain data;
- Artificial neural networks to facilitate multivariate nonlinear regression,
ordination and c1ustering, multivariate time series analysis, image analysis at
micro and macro scale;
- Genetic and evolutionary algorithms for the discovery and evolving of
multivariate nonIinear ruIes, functions, differential equations and artificial neural
networks; - Hybrid and AI models by the embodiment of evolutionary algorithms
in process-based differential equations, the embodiment of fuzzy logic in artificial
neural networks or knowledge processing;
- Adaptive agents to facilitate adaptive simulation and prediction of ecosystem
composition and evolution.
The present book is an outcome of the International Conference on
Applications of Machine Learning to Ecological Modelling, 27 November to 1
December 2000, Adelaide, Australia, which conc1uded with the foundation of the
International
Society
for
Ecological
Informatics
(ISEI)
(http://www.waite.adelaide.edu.auJISEI/). The chapters of the present book are
based on selected papers of the conference, which are exemplary for current
research trends in ecological informatics.
Chapters 1 to 5 address principles and ecological application of fuzzy logic,
artificial neural networks, genetic algorithms, evolutionary computation and
adaptive agents. Salski summarizes concepts of fuzzy logic and discusses
applications for knowledge-based modeling, c1ustering and kriging related to
ecotoxicological, geological and population dynamics data. Giraudel and Lek
discuss the design and appIication of unsupervised artificial neural networks for
the c1assification and visualization of multivariate ecological data. They
demonstrate the potential of Kohonen-type algorithms by c1ustering data of forest
communities in Wisconsin (USA). Morrall discusses origins and nature of genetic
algorithms, and their suitability to induce numericalor rule-based models for
ecological applications. Whigham and Fogel provide a scope of evolutionary
algorithms and their potential for evolving rules, algebraic and differential
equations relevant to ecology. They also address developments on individual and
cooperative behaviour, prey-predator algorithms and hierarchical ecosystems
based on evolutionary algorithms. Recknagel reflects on HoIland's adaptive agents
concept and its potential to more realistically simulate emergent ecosystem
structures and behaviours. He distinguishes between individual-based and state
variable-based agents, and emphasizes on the embodiment of evolutionary
computation in state-variable based agents.
Chapters 6 to 9 provide case studies for the prediction and elucidation of stream
ecosystems by means of machine learning techniques. Goethals, Dedecker,
Gabriels and de Pauw demonstrate applications of c1assification trees and artificial