Toward machine emotional intelligence: Analysis of affective physiologicalstate

The ability to recognize emotion is one of the hallmarks of emotional intel ligence, an aspect of human intelligence that has been argued to be even mo re important than mathematical and verbal intelligences. This paper propose s that machine intelligence needs to include emotional intelligence and dem onstrates results toward this goal: developing a machine's ability to recog nize human affective state given four physiological signals. We describe di fficult Issues unique to obtaining reliable affective data and collect a la rge set of data from a subject trying to elicit and experience each of eigh t emotional states, daily, over multiple weeks. This paper presents and com pares multiple algorithms for feature-based recognition of emotional state from this data. We analyze four physiological signals that exhibit problema tic day-to-day variations: The features of different emotions on the same d ay tend to cluster more tightly than do the features of the same emotion on different days. To handle the daily variations, we propose new features an d algorithms and compare their performance, We find that the technique of s eeding a Fisher Projection with the results of Sequential Floating Forward Search improves the performance of the Fisher Projection and provides the h ighest recognition rates reported to date for classification of affect from physiology: 81 percent recognition accuracy on eight classes of emotion, i ncluding neutral.