Fusion of soft and hard computing techniques: a multi-dimensional categorization scheme

Over the next decade, the fusion of soft computing and hard computing will play a very important role in the development of computationally intelligent systems for various application areas. Up to now, the interest in this field has mainly been driven by successful fusion applications. Researchers still focus on either soft or hard computing techniques. Therefore, it is imperative to treat the emerging field more systematically. This article introduces a multidimensional categorization scheme for fusion techniques that covers various fusion aspects: the degree of interconnection of soft and hard computing components, the topology of fusion techniques, the time when fusion takes place, the layer of a system architecture where fusion takes place, and finally the motivation for the application of fusion techniques. These criteria help to facilitate the discussion of advantages or drawbacks of certain fusion approaches and support the development of novel fusion techniques. The use of these criteria is illustrated by means of some concrete fusion techniques.