ACII 2021 Special Session

Title: Ordinal Affective Computing

Organizers: Georgios N. Yannakakis, Carlos Busso, Shri Narayanan, and Roddy Cowie 


Psychological theories and evidence from multiple disciplines including neuroscience, economics and artificial intelligence suggest that the task of assigning reference-based values to subjective notions is better aligned with the underlying representations. An increasing number of studies have reported the advantages of ordinal annotation over alternative methods (e.g., nominal and interval descriptors) with respect to both reliability and validity. The impact of such evidence on affective computing is extremely important in the ways we annotate, analyze and process emotions. The emerging approach of using ordinal representations has led to improved performance across several tasks, including face analysis, speech recognition, body-based affective interaction, game applications, and retrieval of music and sounds.

Studies relying on the ordinal nature of emotions are scattered across different venues, including several conferences related to affective computing (ACM ICMI, IEEE FG, Interspeech, ICASSP, ACII). The second ACII special session on the topic (the first was held in ACII 2019) has the unique opportunity to bring together researchers working on, but not limited to:

– Psychological methods and tools for the ordinal representation of emotions 

– Statistical methods for ordinal label analysis 
– Preference learning and ranking-based methods for emotion recognition 

– Ordinal methods for the annotation of emotional behaviours 
– Ordinal multimodal corpora 
– Software for ordinal label processing 

This special session aims to further increase the awareness of the ACII community about the advantages of ordinal representations equipping both senior and young scientists with theoretical frameworks, methods and tools and, hence, reframing their working practices from an ordinal perspective