A novel hybrid support vector machine with firebug swarm optimization

In the light of the Industry 4.0 revolution, we find ourselves contending with datasets characterized by a multitude of features, thereby introducing a novel challenge in effectively managing high-featured data in real time. To address this challenge, we propose a new hybrid classification model that leverages the support vector machine (SVM) and firebug swarm optimization (FSO). SVM stands out as a fundamental and widely used classification technique in data science. The optimal determination of kernel parameters and the selection of the most impactful features significantly influence SVM performance. Additionally, the FSO algorithm is noteworthy for its utilization of element-wise Hadamard matrix multiplication operations to update positions. This operation enables parallel execution on multiple data items, thereby reducing overall execution time. Given the ability of FSO algorithm to run on multi-core architectures, we introduce a binary FSO designed for the selection of effective features with the aim of reducing computation time for high-featured data. Our proposed model undergoes evaluation using fifteen UCI datasets and is compared against well-known metaheuristic algorithms that have historically enhanced SVM performance. The experimental results demonstrate that our innovative approach, referred to as FSO–SVM, not only enhances classification accuracy, especially in scenarios with large feature sets but also identifies the smallest subset of features tailored to specific problems. Furthermore, our model characterized through a time-parsimonious analysis, exhibits fast and accurate predictions for high-featured data substantiated by statistical analysis.