Are Instructed Emotional States Suitable for Classification? Demonstration of How They Can Significantly Influence the Classification Result in An Automated Recognition System

TitleAre Instructed Emotional States Suitable for Classification? Demonstration of How They Can Significantly Influence the Classification Result in An Automated Recognition System
Publication TypeJournal Article
Year of PublicationIn Press
AuthorsMagdin, M., and F. Prikler
JournalInternational Journal of Interactive Multimedia and Artificial Intelligence
ISSN1989-1660
IssueIn Press
VolumeIn Press
NumberIn Press
Date Published03/2018
Pagination1-7
Abstract

At the present time, various freely available or commercial solutions are used to classify the subject's emotional state. Classification of the emotional state helps us to understand how the subject feels and what he is experiencing in a particular situation. Classification of the emotional state can thus be used in various areas of our life from neuromarketing, through the automotive industry (determining how emotions affect driving), to implementing such a system into the learning process. The learning process, which is the (mutual) interaction between the teacher and the learner, is an interesting area in which individual emotional states can be explored. In this pedagogical-psychological area several research studies were realized. These studies in some cases demonstrated the important impact of the emotional state on the results of the students. However, for comparison and unambiguous classification of the emotional state most of these studies used the instructed (even constructed) stereotypical facial expressions of the most well-known test databases (Jaffe is a typical example). Such facial expressions are highly standardized, and the software can recognize them with a fairly big percentage, but this does not necessarily point to the actual success rate of the subject's emotional classification in such a test because the similarity to real emotional expression remains unknown. Therefore, we examined facial expressions in real situations. We have subsequently compared these examined facial expressions with the instructed expressions of the same emotions (the Jaffe database). The overall average classification score in real facial expressions was 94.58%.

KeywordsEmotion, Face Detection, Facial Expression, Instructed Stimuli, Real-life Stimuli
DOI10.9781/ijimai.2018.03.002
URLhttp://www.ijimai.org/journal/sites/default/files/files/2018/03/ip_03_02_pdf_42961.pdf
AttachmentSize
ip_03_02.pdf2.01 MB