Predicting Trait Impressions of Faces Using Classifier Ensembles

摘要: In the experiments presented in this chapter, single classifier systems and ensembles are trained to detect social meanings people perceive facial morphology. Exploring machine models of people’s impressions faces has value fields psychology human-computer interaction. Our first concern designing study was developing a sound ground truth for problem domain. We accomplished by collecting large number that exhibited strong human consensus comprehensive set trait categories. Several ensemble composed Levenberg-Marquardt neural networks using different methods collaboration were then match perception six dimensions intelligence, maturity, warmth, sociality, dominance, trustworthiness. results show learning employing as capable most individual beings their ability predict certain make on average observer. Single did not performance well did. Included chapter is tutorial, suitable novice, collaborative used reported study.