A multimodal sine-cosine framework for feature selection applications

Abstract

Not all data features are crucial for uncovering hidden knowledge within various datasets, making the reduction of their dimensional attributes a significant area of interest. This article introduces a novel approach to feature selection within the domain of the sine-cosine algorithm, employing a multimodal optimization strategy. The method is integrated into a wrapper feature selection model that incorporates a range of evaluation function algorithms tied to classifier error rates. The process comprises two key stages: first, the multimodal sine-cosine algorithm is utilized during the feature selection phase, followed by the classification of potential outcomes through the sine-cosine algorithm with the ENN (extended nearest neighbor) method. The suggested technique was assessed through the average number of selected features, average classification accuracy, and best fitness. Also, we compared the performance of the proposed algorithm with some multimodal and non-multimodal optimization techniques. Upon experiments and comparisons, it was evident that the proposed multimodal sine-cosine method yielded the most superior results across all the datasets evaluated from the UCI machine learning repository. Therefore, the utilization of this algorithm for pattern classification was proven to be effective in enhancing classification performance.