Soft Computing is all about inexact solutions for computationally difficult tasks, such as using polynomial time approximation algorithms for NP-complete problems. Soft Computing can be considered a paradigm shift away from the more traditional ‘hard computing’ perception, where data and results are precise. In some respects, Soft Computing is influenced by and based on the human mind, which has a remarkable ability to identify and store relevant information among copious amounts of information, much of which may be vague and only partially observed or sensed.
Soft Computing is already playing a significant role in delivering computation solutions in science and engineering applications. Many Soft Computing techniques are suited to a wide variety of applications, for which the data and inputs that are worked on may be of varying levels of uncertainty and partial truths. This may be extremely important for safety-critical systems (e.g. a ballistic missile guidance control system), in which the system needs to continue working as reliably as possible, even when data may be patchy and sensors working erratically.
The Soft Computing Research Group at the UCT Department of Electrical Engineering provides postgraduate students with the opportunity to be involved with research projects related to Soft Computing. Research in this group is divided into the following three main areas:
Fuzzy Logic: a form of multi-valued logic, in which truth values are considered to be on the continuum [0-1] as opposed to the Boolean 0 or 1 options.
Neural Computing: this involves developing artificial neural networks (or ANNs), which are mathematical computer-based simulations that attempt to mimic the operation of biological neural networks (i..e. as in how the human brain works).
Evolutionary Computation: including evolutionary algorithms and harmony search, swarm intelligence, probability chaos theory and learning theory.
Postgraduate students completing their specialisation in this field will find a number of career possibilities in the Soft Computing field. The demand for experts in this area is high across many sectors in different countries, ranging from consumer products, manufacturing and industrial applications to highly complex, high-risk space systems.