Exploring the Time-efficient Evolutionary-based Feature Selection Algorithms for Speech Data under Stressful Work Condition

Initially, the goal of Machine Learning (ML) advancements is faster computation time and lower resources, while curse dimensionality burdens both resource. This paper describes benefits Feature Selection Algorithms (FSA) for speech data under workload stress. FSA contributes to reducing dimension simultaneously retains information. We chose use robust Evolutionary Algorithm, Harmony Search, Principal Component Analysis, Genetic Particle Swarm Optimization, Ant Colony Bee which are then be evaluated using hierarchical machine learning models. These FSAs explored with conversational stress a Customer Service hotline, has daily complaints that trigger in speaking. Furthermore, we employed precisely 223 acoustic-based features. Using Random Forest, our evaluation result showed had improved 3.6 than original features employed. Evaluation Support Vector beat record 0.001 seconds time.