Developing an understanding of soft computing paradigms in data mining

Abstract

In traditional Boolean computing, the employed logic classifies things in a binary fashion – into absolute truths (1s) or absolute fallacies (0s). However, there are a multitude of real-life scenarios wherein a binary classification would be rendered meaningless. From political orientation to simple colour coding scheme – there are situations in which the outcomes can take a range of values and therefore, can’t be accounted for using Boolean logic. Fuzzy logic, on the other hand, accounts for the possibility of multiple or partial truths lying between the extremes of absolute truths (1) and absolute fallacies (0). The extent of the truth depends upon the proximity of the value to the extremes. This formulation is also very helpful because it resembles the way in which our brain processes the information. We tend to have a prior, which is a set of beliefs constituted of partial truths, and our exposure to new information leads to multiple iterations updating these partial truths values. When these values cross a threshold, that is when these beliefs translate into concrete actions and mental constructs such as opinions, orientation etc. It is also useful because it helps us handle vagueness and inaccuracies which surround a lot of real-life scenarios and are difficult to incorporate in a two-variable setting used by Boolean logic.