Open AccessTraining Affective Computer Vision Models by Crowdsourcing Soft-Target Labels
Author(s) -
Peter Washington,
Haik Kalantarian,
Jack Kent,
Arman Husic,
Aaron Kline,
Émilie Leblanc,
Cathy Hou,
Cezmi Mutlu,
Kaitlyn Dunlap,
Yordan Penev,
Nate Stockham,
Brianna Chrisman,
Kelley Paskov,
Jae-Yoon Jung,
Catalin Voss,
Nick Haber,
Dennis P. Wall
Publication year - 2021
Publication title -
cognitive computation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.86
H-Index - 52
eISSN - 1866-9964
pISSN - 1866-9956
DOI - 10.1007/s12559-021-09936-4
Subject(s) - crowdsourcing , computer science , artificial intelligence , softmax function , classifier (uml) , pattern recognition (psychology) , facial expression , support vector machine , artificial neural network , machine learning , natural language processing , world wide web
Emotion detection classifiers traditionally predict discrete emotions. However, emotion expressions are often subjective, thus requiring a method to handle compound and ambiguous labels. We explore the feasibility of using crowdsourcing to acquire reliable soft-target labels and evaluate an emotion detection classifier trained with these labels. We hypothesize that training with labels that are representative of the diversity of human interpretation of an image will result in predictions that are similarly representative on a disjoint test set. We also hypothesize that crowdsourcing can generate distributions which mirror those generated in a lab setting.