Bridging the Gap between Machine Learning and Affective Computing

Book excerpt: Affective computing refers to computing that relates to, arises from, or influences emotions, as pioneered by Rosalind Picard in 1995. The goal of affective computing is to bridge the gap between human and machines and ultimately enable robots to communicate with human naturally and emotionally. Recently, the research on affective computing has gained considerable progress with many fields contributing including neuroscience, psychology, education, medicine, behavior, sociology, and computer science. Current research in affective computing mainly focuses on estimating of human emotions through different forms of signals, e.g., face video, EEG, Speech, PET scans or fMRI. Inferring the emotion of humans is difficult, as emotion is a subjective, unconscious experience characterized primarily by psycho-physiological expressions and biological reactions. It is influenced by hormones and neurotransmitters such as dopamine, noradrenaline, serotonin, oxytocin, GABA… etc. The physiology of emotion is closely linked to arousal of the nervous system with various states and strengths relating, apparently, to different particular emotions. To understand “emotion” or “affect” merely by machine learning or big data analysis is not enough, but the understanding and applications from the intrinsic features of emotions from the neuroscience aspect is essential.