TY - GEN
T1 - Investigating word affect features and fusion of probabilistic predictions incorporating uncertainty in AVEC 2017
AU - Dang, Ting
AU - Stasak, Brian
AU - Huang, Zhaocheng
AU - Jayawardena, Sadari
AU - Atcheson, Mia
AU - Hayat, Munawar
AU - Le, Phu
AU - Sethu, Vidhyasaharan
AU - Goecke, Roland
AU - Epps, Julien
PY - 2017/10/23
Y1 - 2017/10/23
N2 - Predicting emotion intensity and severity of depression are both challenging and important problems within the broader field of affective computing. As part of the AVEC 2017, we developed a number of systems to accomplish these tasks. In particular, word affect features, which derive human affect ratings (e.g. arousal and valence) from transcripts, were investigated for predicting depression severity and liking, showing great promise. A simple system based on the word affect features achieved an RMSE of 6.02 on the test set, yielding a relative improvement of 13.6% over the baseline. For the emotion prediction sub-challenge, we investigated multimodal fusion, which incorporated a measure of uncertainty associated with each prediction within an Output-Associative fusion framework for arousal and valence prediction, whilst liking prediction systems mainly focused on text-based features. Our best emotion prediction systems provided significant relative improvements over the baseline on the test set of 39.5%, 17.6%, and 29.3% for arousal, valence, and liking. Of particular note is that consistent improvements were observed when incorporating prediction uncertainty across various system configurations for predicting arousal and valence, suggesting the importance of taking into consideration prediction uncertainty for fusion and more broadly the advantages of probabilistic predictions.
AB - Predicting emotion intensity and severity of depression are both challenging and important problems within the broader field of affective computing. As part of the AVEC 2017, we developed a number of systems to accomplish these tasks. In particular, word affect features, which derive human affect ratings (e.g. arousal and valence) from transcripts, were investigated for predicting depression severity and liking, showing great promise. A simple system based on the word affect features achieved an RMSE of 6.02 on the test set, yielding a relative improvement of 13.6% over the baseline. For the emotion prediction sub-challenge, we investigated multimodal fusion, which incorporated a measure of uncertainty associated with each prediction within an Output-Associative fusion framework for arousal and valence prediction, whilst liking prediction systems mainly focused on text-based features. Our best emotion prediction systems provided significant relative improvements over the baseline on the test set of 39.5%, 17.6%, and 29.3% for arousal, valence, and liking. Of particular note is that consistent improvements were observed when incorporating prediction uncertainty across various system configurations for predicting arousal and valence, suggesting the importance of taking into consideration prediction uncertainty for fusion and more broadly the advantages of probabilistic predictions.
KW - Depression prediction
KW - Dimensional emotion prediction
KW - Output-associative fusion
KW - Sentiment analysis
KW - Uncertainty prediction
KW - Outputassociative fusion
UR - http://www.scopus.com/inward/record.url?scp=85035314372&partnerID=8YFLogxK
UR - http://www.mendeley.com/research/investigating-word-affect-features-fusion-probabilistic-predictions-incorporating-uncertainty-avec-2
U2 - 10.1145/3133944.3133952
DO - 10.1145/3133944.3133952
M3 - Conference contribution
AN - SCOPUS:85035314372
T3 - AVEC 2017 - Proceedings of the 7th Annual Workshop on Audio/Visual Emotion Challenge, co-located with MM 2017
SP - 27
EP - 35
BT - AVEC 2017 - Proceedings of the 7th Annual Workshop on Audio/Visual Emotion Challenge, co-located with MM 2017
PB - Association for Computing Machinery (ACM)
CY - Mountain View, USA
T2 - 7th Annual Workshop on Audio/Visual Emotion Challenge, AVEC 2017
Y2 - 23 October 2017
ER -