Automatic instance generation using simulation for inductive learning

Inductive learning can be used to extract rules required for an expert system which assists in output analysis for system simulation. However, several examples of the system, constituting an instance set, are required for learning to take place. Generating the required instance set to be used by-an inductive learning algorithm is time consuming and complex. This paper is an attempt to clarify this problem, discuss its complexity and suggest context related solutions. A procedure for automatic instance generation is then proposed. The proposed procedure is a combination of three search methods (grid based, forward search ,backward search).