Computational intelligence and soft computing: some thoughts on already explored and not yet explored paths

Abstract

We comment upon the very essence, roots, potentials, and applicability of computational intelligence and soft computing. We followed a different path than those traditionally employed, and which are so well and in a deep and comprehensive way documented in other papers in this special issue. First, we consider relations between computational intelligence and artificial intelligence, starting from a reference to different kinds of views if how intelligence is meant in science, and whether it has a general nature or many different types. Then, we consider the issue of symbolic and numerical calculations, and the two attitudes for problem solving: a general one and problem specific. Then, we discuss soft computing which is in our view a narrower area than computational intelligence and consider two aspects implied by the inclusion of the two words: "computing" and "soft". First, we discuss how soft computing relates to the traditionally meant computing [in the sense of (a theory of) computation]. Second, we pursue a novel path that has not been practically considered in the literature, that is – first - relations of soft computing to soft sciences, and – second – relations of soft computing to Checkland’s soft systems methodology. We hope that this paper will trigger a discussion and stimulate a new line of research with both soft computing and computational intelligence communities, on the one hand, and a broadly perceived soft science communities, on the other hand, that can bring about new ideas and novel problem formulations and solutions.