A Novel Experiment Setting for Cross-subject Emotion Recognition.

Area of Science:

• Affective Computing
• Human-Computer Interaction
• Machine Learning

Background:

• Cross-subject emotion recognition is crucial for developing adaptive technologies.
• Traditional experiments use identical video stimuli, potentially introducing shared noise and classification bias.
• Existing models struggle to mitigate the impact of uniform video exposure on recognition accuracy.

Purpose of the Study:

• To propose a novel experimental setting for cross-subject emotion recognition.
• To address the limitations of traditional experiment designs in emotion recognition research.
• To evaluate the effectiveness of a new experimental approach compared to existing methods.

Main Methods:

• Evaluation of different experimental settings for cross-subject emotion recognition.
• Utilizing four public emotion datasets: DEAP, SEED, SEED-IV, and SEED-V.
• Comparative analysis of traditional versus proposed experimental methodologies.

Main Results:

• Traditional experiment settings exhibit deficiencies in cross-subject emotion recognition.
• The proposed novel experiment setting demonstrates significant advantages.
• The new approach effectively mitigates bias introduced by identical video stimuli.

Conclusions:

• The traditional experimental setup for cross-subject emotion recognition is suboptimal.
• The proposed experiment setting offers a more robust and accurate methodology.
• This work advances the field of emotion recognition by refining experimental design.