A Novel Feature Selection Method Based on Slime Mold Network Formation Behavior

Slime mold (Physarum polycephalum) is a remarkable organism that can solve complex problems such as mazes, shortest path networks, and optimal transport networks. Inspired by the adaptive network formation behavior of slime mold, we propose a novel feature selection algorithm (SMFS). The SMFS converts the feature selection into an optimal subgraph problem and employs a slime mold network formation inspired strategy to guide the sub-graph search procedure. We evaluate the performance of SMFS against 4 well-known meta-heuristic feature selection methods using different classifiers (support vector machine and naive Bayes). The experimental results on several benchmark datasets demonstrate the efficiency and effectiveness of the SMFS method and its superiority over previous related methods.