Starting with the definition of pattern recognition and soft computing, their relevance to machine intelligence and data mining is explained. Different synergistic integrations of various soft computing technologies are stated. The role of rough sets and rough-fuzzy computing in uncertainty handling and granular computing is highlighted along with their merits. Concepts of rough information granules and their applications are stated. Example problems considered in rough-fuzzy framework include: clustering, case selection, knowledge encoding in granular neural networks, bio-bases selection from protein sequences, measuring image entropy and video tracking. Superiority of the integrated paradigm in mining is demonstrated on real life problems together with the significance of neighbourhood rough sets. Certain challenging issues involved with granular computing are enumerated. The talk concludes mentioning their relevance to computational theory of perception, natural computing and handling big data.