Soft computing is completely opposite to traditional computing. It deals with approximate models and gives solutions to complex real-life problems. Unlike hard computing, soft computing is tolerant of imprecision, approximations, and uncertainty. Soft computing is based on soft computing techniques i.e it provides rapid dissemination in evolutionary algorithms and genetic programming, neural science and neural net systems, fuzzy set theory and fuzzy systems, and chaos theory and chaotic systems. In one line we can say that soft computing finds the solution of NP-complete problems for which there is no algorithm that can compute an exact solution in polynomial time.
One of the problems with traditional computing is that complex plants cannot be accurately described by the mathematical models, and are therefore difficult to control using such existing methods. The guiding principle of soft computing is to exploit the tolerance for uncertainty to achieve the robustness, better rapport with reality. It also helps in non-linear programming. Over the years, it has become very popular and has drawn research interest from people with different backgrounds.
Many non-linear plants with large time delays cannot be easily controlled and stabilized using tradition computing techniques. Soft computing provides an efficient way of handling such plants. If we talk about soft computing in a more specific way then it is not a single term but several other terms are associated with it. It contains many methods such as fuzzy logic, neural networks, and genetic algorithms. these methods are used in complement to each other. Here is a brief about each on three:
1. Fuzzy logic
The fuzzy logic helps in transforming the crisp input value into a fuzzy linguistic value. It works simply on the basis "IF-THEN" rules involving linguistic variables. The output actions produced by the process of defuzzification. This process provides the design of non-linear control systems which are difficult to handle using traditional methods.
2.  Artificial neural networks
Artificial neural networks(ANN), or simply neural computing is one of the rapidly growing fields and attracting engineering disciplines like electronic engineering, control engineering, and software engineering. A neural network is a large network of interconnected elements inspired by human neurons. This network has to be trained so that a known set of inputs produces the desired outputs. Teaching patterns are fed to the network and let the network to change its weighting function according to some previously defined learning rules.
(a) ANNs are not universal tools for solving the problems.
(b) ANN can deal with incomplete data sets.
(c)The result of ANN depends entirely on the accuracy of data.
(d) ANN is most useful in prediction applications
3. Genetic algorithms
They are parts of artificial intelligence and fuzzy computing. The basic principle is to mimic the natural selection in nature in order to find a good selection for an application. Genetic algorithms can be used for finding solutions to complex search problems found in engineering applications. They work on the method that they search through various designs and components to find the best combination that will result in overall better and cheaper design.
As the power of computing devices increasing these days, soft computing is becoming more important. Nowadays, most soft computing applications can be handled efficiently by low-cost but super-fast microcontrollers. You will be surprised to know that all your daily use electronic appliances such as washing machine, fridge, all use concepts of fuzzy logic, artificial neural networks. And that's why soft computing is expected to grow within the next decade together with the use of IoT(Internet of things) in future, domestic, industrial and commercial markets.