An Integrated Approach to Speed Up GA-SVM Feature Selection Model

Significant information or features are often overshadowed by noises and resulted in poor classification results. Feature selection methods such as GA-SVM are desirable in filtering out the irrelevant features and thus improve the accuracy; the selection itself might also offer critical insights into the problems. However, the high computational cost greatly discourages the application of GA-SVM, especially for large-scale datasets. In this paper, an HPC-enabled GA-SVM (HGA-SVM) is proposed and implemented by integrating data parallelization, multithreading and heuristic techniques with the ultimate goal of maintaining robustness and lowering computational cost. Our proposed model is comprised of four improvement strategies: 1) GA Parallelization, 2) SVM Parallelization, 3) Neighbor Search and 4) Evaluation Caching. All the four strategies improve the respective aspects of the feature selection algorithm and contribute collectively towards higher computational throughput.