Uncovering Cross-Cultural Emotional Patterns: t-SNE and PCA on Micro-Expressions

Facial Expression Recognition (FER) has advanced significantly in recent years, yet its generalizability remains a challenge, as existing FER methods often overlook ethnic variations in emotional expression. Micro-expressions are brief, involuntary facial movements that reveal genuine emotions, making them particularly useful for understanding subtle contrasts across cultural contexts. In this study, we analyze cultural differences in micro-expressions by integrating data from two publicly available datasets; with Southern European and Eastern Asian participants sourced from 4DME, and White British participants from SAMM. The integrated dataset, SAMM+4DME, comprises five common features extracted using ethnicity annotation and features matching via filtering and AU standardization techniques. Visualization of dimensionality reduced features using t-SNE and PCA reveal distinct clusters that highlight significant variations in micro-expression patterns across ethnic groups. This finding emphasizes the necessity of culture-aware FER systems to enhance the generalizability of micro-expression recognition across diverse populations.