Inference of personality traits and affect schedule by analysis of spontaneous reactions to affective videos

摘要: This paper presents a method for inferring the Positive and Negative Affect Schedule (PANAS) BigFive personality traits of 35 participants through analysis their implicit responses to 16 emotional videos. The employed modalities record are (i) EEG, (ii) peripheral physiological signals (ECG, GSR), (iii) facial landmark trajectories. predictions traits/PANAS done using linear regression models that trained independently on each modality. main findings this study that: PANAS individuals can be predicted based users' affective video content, ECG+GSR yield 70%±8% F1-score distinction between extroverts/introverts, EEG 69%±6% creative/non creative people, finally (iv) prediction agreeableness, stability, baseline states we achieved significantly higher than chance-level results.