Simple Heuristics for the Choquet Integral Classifier
Ken Adams · 2015
The Choquet Integral is a successful classification method. However, like other methods, when applied to large data sets where many coefficients have to be optimised, search methods such as genetic algorithms (GA) are used. In this paper, heuristics are developed that can achieve reasonably accurate results by using information gained from the data set. This enables optimisation using a smaller set of coefficients than those needed in the original search space and this may provide a good starting place for a GA search and other optimisation methods. For the purposes of this research, the data used is the Wisconsin Breast Cancer data set, and it is envisaged that the methods described here will generalise onwards to many other optimisation techniques and data sets.