Machine learning based classification of presence utilizing psychophysiological signals in immersive virtual...

Area of Science:

• Neuroscience
• Computer Science
• Human-Computer Interaction

Background:

• User presence in virtual reality (VR) enhances engagement.
• Current presence measurement methods, like questionnaires, are subjective.
• Objective, data-driven methods are needed to quantify VR presence.

Purpose of the Study:

• To develop a machine learning model for detecting varying levels of VR presence.
• To utilize multimodal neurological and physiological signals for presence detection.
• To compare the efficacy of various machine learning classifiers for this task.

Main Methods:

• An experiment with 22 participants exposed to high, medium, and low VR presence levels.
• Systematic manipulation of graphical fidelity, audio cues, latency, and haptic feedback.
• Utilizing electroencephalography and electrodermal activity signals with machine learning classifiers (SVM, KNN, XGBoost, RF, LR, MLP).

Main Results:

• The Multiple Layer Perceptron (MLP) model achieved the highest classification accuracy.
• Relative band power (e.g., beta/theta ratio) and differential entropy in frontal and parietal regions were key indicators.
• Neurological and physiological signals effectively differentiated presence levels.

Conclusions:

• Machine learning models can objectively detect presence levels in VR.
• MLP demonstrates strong performance in classifying VR presence using biosignals.
• Brain and physiological activity patterns are reliable markers for VR presence.