C18 - Identification of Nonlinear Dynamics – Neural Networks versus Transfer Functions
Benchmark C18 Identification of Nonlinear Dynamics – Neural Networks versus Transfer Functions studies alternative approaches for identification of the nonlinear dynamical relation between muscle force and muscle-belly thickening. Classical discrete transfer function models and as alternatives discrete transfer function models with neural net models in parallel, and extended neural net models are to be compared with respect to accuracy and efficiency. Two data sets, measured on the same muscle type, are available for identification procedures and for validation purposes (download).
The comparative approaches invite for educational use in modelling of nonlinear dynamics and in introduction to neural nets – independent on the application area physiology. Another educational aspect is use of appropriate discrete transfer functions and error correction by neural network compensation.
Modellers and simulationists are invited to prepare, to realise, and to submit a
- C18 Benchmark Solution with concise description of model implementation and experimentation tasks (two pages SNE), or preferably a
- C18 Benchmark Report with sufficient detailed description of model implementation with variants and adequate experiment formulations (four to six pages SNE).