Multitask Emotion Recognition with Incomplete Labels
This addresses the practical problem of incomplete labeling in emotion recognition datasets, though it appears incremental as it builds on existing multitask learning approaches.
The authors tackled multitask emotion recognition with incomplete labels by proposing a teacher-student algorithm that learns from missing labels and applying data balancing techniques, achieving performance improvements where most student models outperformed their teachers on all three tasks.
We train a unified model to perform three tasks: facial action unit detection, expression classification, and valence-arousal estimation. We address two main challenges of learning the three tasks. First, most existing datasets are highly imbalanced. Second, most existing datasets do not contain labels for all three tasks. To tackle the first challenge, we apply data balancing techniques to experimental datasets. To tackle the second challenge, we propose an algorithm for the multitask model to learn from missing (incomplete) labels. This algorithm has two steps. We first train a teacher model to perform all three tasks, where each instance is trained by the ground truth label of its corresponding task. Secondly, we refer to the outputs of the teacher model as the soft labels. We use the soft labels and the ground truth to train the student model. We find that most of the student models outperform their teacher model on all the three tasks. Finally, we use model ensembling to boost performance further on the three tasks.