An Efficient Algorithm for Finding a Fuzzy Rough Set Reduct Using an Improved Harmony Search

To increase learning accuracy, it is important to remove misleading, redundant, and irrelevant features. Fuzzy rough set offers formal mathematical tools to reduce the number of attributes and determine the minimal subset. Unfortunately, using the formal approach is time consuming, particularly if a large dataset is used. In this paper, an efficient algorithm for finding a reduct is introduced. Several techniques are proposed and combined with the harmony search, such as using a balanced fitness function, fusing the classical ranking methods with the fuzzy-rough method, and applying binary operations to speed up implementation. Comprehensive experiments on 18 datasets demonstrate the efficiency of using the suggested algorithm and show that the new algorithm outperforms several well-known algorithms.