Structural modeling using rough sets

The goal of structural modeling is to determine the strengths of relationships between variables representing a system. Applications done by the authors show that the use of the rough sets has a higher processing accuracy compared to pure statistical approaches. Intelligent inferences from upper approximation of the rough sets explain the advantage. Statistics may blur the results of data processing by selecting sample elements at random. The rough sets use the upper and lower approximations to identify and utilize contexts of each specific object in available data and reveal casual relationships between objects. The upper approximation adds specific information to each object in question from the context, thanks to which relationships between objects become sharpened. Examples are presented which illustrate that image filters based on statistics blur image details, whereas filters based on rough sets enhance and sharpen the image. A new neural network is described and applied as a structural equation system for uncovering casual relationships using a rough sets approach