Affective Video Content Analyses by Using Cross-Modal Embedding Learning Features | IEEE Conference Publication | IEEE Xplore