RESEARCH

HOME RESEARCH
Behavior Computing
States and Traits
Spoken Dialogs
Predicting Collaborative Task Performance Using Graph Interlocutor Acoustic Network in Small Group Interaction
Abstract
Recent works have demonstrated that the integration of grouplevel personality and vocal behaviors can provide enhanced prediction power on task performance for small group interactions. In this work, we propose that the impact of member personality for task performance prediction in groups should be explicitly modeled from both intra and inter-group perspectives. Specifically, we propose a Graph Interlocutor Acoustic Network (GIAN) architecture that jointly learns the relationship between vocal behaviors and personality attributes with intra-group attention and inter-group graph convolutional layer. We evaluate our proposed G-IAN on two group interaction databases and achieve 78.4% and 72.2% group performance classification accuracy, which outperforms the baseline model that models vocal behavior only by 14% absolute. Further, our analysis shows that Agreeableness and Conscientiousness demonstrate a clear positive impact in our model that leverages the inter-group personality structure for enhanced task performance prediction.
Figures
A complete schematic of our Graph Interlocutor Acoustic Network (G-IAN). It applies modified attention mechanism controlled by group-level personality, and models the inter-group relationship of personality with a graph convolutional layer for the recognition task.
A complete schematic of our Graph Interlocutor Acoustic Network (G-IAN). It applies modified attention mechanism controlled by group-level personality, and models the inter-group relationship of personality with a graph convolutional layer for the recognition task.
Keywords
group interaction | personality | attention mechanism | graph convolutional network
Authors
Publication Date
2020/10/25
Conference
Interspeech
Interspeech 2020
DOI
10.21437/Interspeech.2020-1698
Publisher
ISCA