Soft computing is a branch of computer science that works on the basis that not all solutions to problems can be precisely accurate. It is most commonly associated with computing techniques that are designed to mimic biology, most notably the human brain. Most problems tackled by soft computing cannot easily be broken down into a purely mathematical approach.
To understand the concept of soft computing, it is necessary to understand the differences between a computer and the human brain, particularly their relative strengths and advantages. The brain works more slowly in carrying out a specific task but is much more skilled at considering multiple options at the same time. Computers can calculate more quickly but are restricted to a more logical, one thing and a time approach.
To give examples of these differences in practice, a search engine can look for a particular piece of text across the entire indexed World Wide Web in a fraction of a second. A human might not be able to complete the same task with the equivalent amount of printed material in a lifetime. Computers are, however, relatively poor at recognizing an image, for example a face. A human can usually recognize a known face in an instant, while even recognizing somebody he met once a long time ago is possible within a few seconds.
This disparity in facial recognition ability is thought to be because humans do a good job of remembering a face as a whole, unlike a computer, which would break an image down into individual pixels and compare them one by one. Meanwhile, the human would be confident in noting enough similarities to be confident about making a strong guess, even if there were some minor or even major differences. A human can usually recognize the face of an old school friend, even if it has changed dramatically through aging; the human does a good job of identifying the features that matter, such as eyes and bone structure.
Soft computing aims to emulate the human, or other animal, approach to tackling problems. This can include the use of fuzzy logic, which is a contrast to traditional binary logic where every piece of data is either a 1 or a 0, which can be thought of in terms of a flat wrong or right. Fuzzy logic allows for a piece of data to be rated at any stage between 0 and 1, equivalent to having infinite degrees of accuracy.
The most common uses of soft computing involve trying to map a biological structure such as the brain. This allows scientists to learn more about how the brain works and how to deal with neurological problems. Soft computing can also be used to make it easier to design software that operates through a logic that humans can understand. It can also be used as the basis of a hybrid approach to computing, combining the reasoning skills of humans with the processing speed and accuracy of a computer.