Toward Emotion Recognition From Early Fused Acoustic and Language Features Using Recursive Neural Networks

Recognising emotions from language is considered an important aspect of affective computing. However, the application of recognised emotions in an effective manner is often bound to the context where the emotion was detected, without the acquisition of information about the relation between spoken words and the recognised emotion. To apply recognised emotions to a broader context, knowledge about this dynamic must be accrued during the emotion classification process. In this paper, we outline a novel method of extracting these relations, using recursive neural networks to process the syntactic structure of speech in order to better understand how emotions are expressed and what spoken words they relate to.

@InProceedings{Sut18, 
 	 author =  {Sutherland, Alexander},  
 	 title = {Toward Emotion Recognition From Early Fused Acoustic and Language Features Using Recursive Neural Networks}, 
 	 booktitle = {International Ph.D. Conference on Safe and Social Robotics (SSR)},
 	 number = {},
 	 volume = {},
 	 pages = {},
 	 year = {2018},
 	 month = {Sep},
 	 publisher = {},
 	 doi = {}, 
 }