The Method for Identifying Employees' Emotions in Adverse States Incorporating PSO-kNN Algorithm and Multiple...

Area of Science:

• Physiological Psychology
• Machine Learning
• Affective Computing

Background:

• Human beings experience a range of negative emotions like fear and boredom.
• These emotions are often associated with varying levels of psychological tension.
• Accurate identification of emotional states is crucial for understanding human well-being.

Purpose of the Study:

• To design an experiment to evoke specific emotions (calm, excited, bored, fearful) and tension levels (low, moderate, high).
• To develop a robust method for classifying these induced emotional and tension states using physiological data.
• To evaluate the effectiveness of a novel feature selection and classification approach.

Main Methods:

• Emotion-evoking experiment to induce target states.
• Physiological signal acquisition (heart rate, respiration rate, etc.).
• Baseline removal preprocessing, Particle Swarm Optimization (PSO) for feature selection, and K-Nearest Neighbors (KNN) classification.

Main Results:

• The proposed method, incorporating baseline removal and PSO feature selection, demonstrated enhanced recognition accuracy.
• Classification performance of the optimized KNN significantly outperformed the traditional KNN algorithm.
• The study successfully classified different emotion and tension levels based on physiological signals.

Conclusions:

• The integration of baseline removal and PSO-based feature selection substantially improves the accuracy of emotion and tension recognition.
• The developed methodology offers a more effective approach for classifying undesirable emotional states using physiological data.
• This research contributes to the advancement of affective computing and emotion recognition technologies.