Case-Based Simulation

Simulation is defined as the experimentation with models. A model for a system and an experiment is anything to which the experiment can be applied in order to answer questions about the system. Often deductive models for systems are difficult to acquire. Case-based and inductive approaches facilitate the acquisition of knowledge and present an alternative which allows for an "easy" application.
 

Fuzzy-Based Inductive Reasoning (1993-1995)

Francois Cellier and his students have been working on the refinement and application of Fuzzy-Based Inductive Reasoning (FIR) since the late 80ties. FIR, methodological rooted in the work by George Klir, requires huge amounts of homogeneously structured cases to derive a behavior and structural model of the system of interest. In collaboration with Francois Cellier and Robert Frye we employed FIR to predict the carbon dioxid dynamic of Biosphere 2.  The project was sponsored by the Feodor Lynen Program of the Alexander von Humboldt
Foundation.
 

OASES - Our Approach to Simulate   Experimental Studies (1995-1998)

Contrary to other case-based approaches which require general knowledge for case matching and adaptation - OASES utilizes the knowledge inherent in cases for case matching and adaptation. Instead of global strategies, local context dependent strategies of matching and adaptation are employed reflecting recent developments in the community of case-based reasoning. Applied in the area of bonehealing OASES has demonstrated its capability to deal with idiosyncrasy. Based on 650 cases out of 50 publications with 70 different variables it predicts the likely result of certain experimental settings in the area of bone healing. Currently an application in the area of Anaesthesiology is under development. The system shall forecast post operational side effects of certain anaesthesiological procedures. The case base comprises 2000 cases out of 200 publications.
The project was sponsored by the Land Baden Württemberg.
 

Publications

For Publications click here.