PRELIMINARY AT A GLANCE | PRELIMINARY FULL
Plenary Talk - Junzo Watada (CV) - Building regression model under hybrid uncertainty
Fuzzy regression model was proposed in early 1980s.
Since then various models have been provided to deal with data under uncertain environment. Most of these models are relating to fuzzy uncertainty. This tutorial will recall these models and discuss their purposes. After then, both uncertainties including randomness and fuzziness are discussed.
In this situation, we will explain related issues in building regression model under such hybrid uncertainty. We provide the model to treat data under such hybrid uncertainty environment.
After then we explain several applications of such a regression model to real usages.
Plenary Talk - Prof. António E. Ruano (CV) - Evolving neural and neuro-fuzzy models for identification, control and biomedical signal processing
Neural Networks and Neuro-fuzzy systems are models that have found a large application spectrum in systems identification, control and signal processing. While there are established techniques for determining their parameter values, the joint determination of their structure, model order and input selection, and model tailoring for the particular application at hand is still an open issue. In this talk we discuss an iterative identification framework, involving multiobjective evolutionary algorithms and gradient based algorithms in an hybrid scheme, that can be used to address the above mentioned problems, for models whose linear and nonlinear parameters can be separated. The use of the proposed framework will be exemplified in cases of systems identification, predictive control and biomedical applications.