Acted vs. Improvised: Domain Adaptation for Elicitation Approaches in Audio-Visual Emotion Recognition
This work addresses the challenge of generalizing emotion recognition systems across varied data collection contexts, but it is incremental as it applies existing domain adaptation techniques to a specific domain issue.
The paper tackled the problem of domain mismatch in audio-visual emotion recognition caused by different emotion elicitation approaches, such as acted vs. improvised sessions, and found that domain transfer learning methods, including gradient reversal and entropy loss, can alleviate this mismatch with limited labeled target samples.
Key challenges in developing generalized automatic emotion recognition systems include scarcity of labeled data and lack of gold-standard references. Even for the cues that are labeled as the same emotion category, the variability of associated expressions can be high depending on the elicitation context e.g., emotion elicited during improvised conversations vs. acted sessions with predefined scripts. In this work, we regard the emotion elicitation approach as domain knowledge, and explore domain transfer learning techniques on emotional utterances collected under different emotion elicitation approaches, particularly with limited labeled target samples. Our emotion recognition model combines the gradient reversal technique with an entropy loss function as well as the softlabel loss, and the experiment results show that domain transfer learning methods can be employed to alleviate the domain mismatch between different elicitation approaches. Our work provides new insights into emotion data collection, particularly the impact of its elicitation strategies, and the importance of domain adaptation in emotion recognition aiming for generalized systems.