An Enhanced Evolutionary Based Feature Selection Approach Using Grey Wolf Optimizer for the Classification of High-dimensional Biological Data

Feature selection (FS) is a pre-processing step that aims to eliminate the redundant and less-informative features enhance performance of data mining techniques. It also considered as one key success factors for classification problems in high-dimensional datasets. This paper proposes an efficient wrapper feature method based on Grey Wolf Optimizer (GWO). GWO recent metaheuristic algorithm has been widely employed solve diverse optimization problems. However, mainly follows search directions toward leading wolves, making it prone fall into local optima, especially when dealing with problems, which case many biological An enhanced variation called EGWO, adapts two enhancements, introduced overcome this specific shortcoming. In first place, transition parameter concept incorporated move from exploration phase exploitation phase. Several adaptive non-linear decreasing formulas are control parameters. second random-based strategy exploited empower diversity during process. Two binarization schemes using S-shaped V-shaped transfer functions map continuous space binary FS. The efficiency proposed EGWO validated ten low-samples data. Our experiments show promising compared original approach other state-of-the-art techniques terms dimensionality reduction enhancement performance.