The computational environment used in any analytical approach is perhaps too categorical and inflexible in order to cope with the intricacy and the complexity of the real world physical systems. It turns out that in dealing with such systems, one has to face a high degree of uncertainty and tolerate imprecision. Soft computing the tolerance for imprecision and uncertainty is exploited to achieve tractability, lower cost, high Machine Intelligence Quotient (MIQ) and economy of communication.
The principal constituents of soft computing are:
1. FUZZY LOGIC (FL) 2. ARTIFICIAL NEURAL NETWORK (ANN) 3. PROBABILISTIC REASONING (PR)
These distinct and yet interrelated methodologies are currently attracting a great deal of attention and have already found a number of practical applications ranging from industrial process control, fault diagnosis and smart appliances to speech recognition and planning under uncertainty. In this perspective, the principal contribution of fuzzy logic relates to its provision of a foundation for approximate reasoning, while neural network theory provides an effective methodology for learning from examples.
