The process of feature selection has a high impact on data mining tasks such as classification and clustering. Removing irrelevant, noisy and redundant data not only increases the quality of the task but also reduces the computational complexity and execution time. Nature-inspired algorithms have tackled the problem of feature selection efficiently. But when applying on a high-dimensional dataset, the metaheuristic algorithms have difficulty to converge. In this paper, an existing artificial bee colony algorithm for feature selection is modified by incorporating a data preprocessing step to reduce the size of the input dataset. The preprocessing step computes the centre of gravity vectors corresponding to the original dataset to form a smaller dataset. The artificial bee colony algorithm works on this smaller dataset for feature selection. The proposed method generates better results with less time and complexity when compared to the existing algorithms.
CITATION STYLE
Bindu, M. G., & Sabu, M. K. (2020). A centre of gravity-based preprocessing approach for feature selection using artificial bee colony algorithm on high-dimensional datasets. In Lecture Notes in Electrical Engineering (Vol. 656, pp. 283–294). Springer. https://doi.org/10.1007/978-981-15-3992-3_23
Mendeley helps you to discover research relevant for your work.