Towards Nature-Inspired Intelligence Search for Optimization of Multi-Dimensional Feature Selection
Khin Sandar Kyaw, Somchai Limsiroratana · 2019
Since news document includes complex and high-dimensional feature, the conventional feature selection scheme becomes inefficient and ineffective for feature engineering in text mining field. Meanwhile, the nature-inspired intelligence (NII) is advanced significantly for solving the complex problem such as multi-dimensional feature selection. NII is the category of non-deterministic that is used the meta-heuristic search capability which consists of a group of search agents for exploring the feasible region based on both randomization and some rules. In this paper, Wolf intelligence-based optimization of multi-dimensional feature selection approach (WI-OMFS) is proposed for news document classification and compared the results of performance to conventional search-based feature selection approach. The performance (accuracy) is used as fitness function. Moreover, various measurements of performance and computation complexity are also used to evaluate the proposed system. According to the experimental results, WI-OMFS can provide robustness for performance according to the objective function.