A reduced variable neighborhood search Aapproach for gene selection in cancer classification

Abstract

In this work we propose a Reduced Variable Neighborhood Search (RVNS) metaheuristic algorithm, to handle the gene selection problem in cancer classification. RVNS is utilized as the primary search method and gene subsets obtained are evaluated by three learning algorithms, namely support vector machine, k-nearest neighbors, and random forest. Experiments are conducted on publicly available cancer-related datasets, all characterized by a small sample size and high dimensionality. Since RVNS seeks gene subsets that yield accurate predictions for all three aforementioned classifiers, the proposed results can be considered more reliable. To the best of our knowledge, our methodology is innovative due to the fact that, it combines the Recursive Feature Elimination (RFE) heuristic with an RVNS algorithm. Despite the large size of the problem instances, the proposed FS scheme converges within reasonably short time. Results indicate high performance of RVNS, which is further improved when the RFE method is applied as a pre-processing step. Moreover, our approach seems to outperform similar recent algorithms in both terms of accuracy and run-time.