LEE- Denizcilik Çalışmaları Lisansüstü Programı

Bu topluluk için Kalıcı Uri

http://hdl.handle.net/11527/19253

Gözat

Konu "machine learning" ile LEE- Denizcilik Çalışmaları Lisansüstü Programı'a göz atma
  • Öge
    Design of an emotion recognition system using machine learning for maritime operations: Development of a cognitive interface with psychophysiological data analysis
    (Graduate School, 2024-06-10) Kordlar Alipanah, Abbas ; Tavacıoğlu, Leyla ; Kamaşak, Mustafa Ersel ; 423211021 ; Maritime Studies
    The maritime sector is experiencing a profound shift towards digitization, which introduces new cyber threats and heightens the impact of human factors. This research develops an emotion recognition system powered by machine learning aimed at monitoring the cognitive states of maritime personnel to address both operational and cybersecurity challenges. The study leverages psychophysiological data including heart rate variability, electrodermal activity, EEG (Electroencephalogram), GSR (Galvanic Skin Response), and eye-tracking systems to classify emotional states accurately. Additionally, self-reported anxiety surveys were utilized to explore the relationship between emotional states and anxiety levels. A detailed analysis of arousal and valence was conducted, categorizing emotions into four quadrants: high arousal positive valence, high arousal negative valence, low arousal negative valence, and low arousal positive valence. This approach enhances the understanding of the emotional states of maritime personnel and their impact on performance and decision-making. The primary objective is to create a precise and dependable emotion recognition model through the application of various machine learning techniques such as Random Forest, SVM, KNN, Logistic Regression, Gradient Boosting, AdaBoost, and LGBMClassifier. The results showed that the LGBMClassifier achieved the highest accuracy of 62%, with an F1-score of 0.60 for class 3 emotions. Feature importance analysis revealed that Alpha, Beta, and Gamma brainwaves, along with HRV and GSR, were significant predictors of emotional states. The findings indicate a clear correlation between emotional states and anxiety levels. Positive emotions are linked to lower anxiety levels, while negative emotions correlate with higher anxiety levels. These insights underline the importance of fostering positive emotional experiences to enhance mental well-being. A comprehensive cognitive interface integrates these findings, enhancing situational awareness and decision-making in real-time. This framework helps recognize and respond to risky states, improving safety, performance, and cybersecurity outcomes. The study's contributions revolutionize emotion management in maritime operations, paving the way for safer, more resilient practices in an increasingly digital and interconnected industry.