Distilling Privileged Multimodal Information for Expression Recognition using Optimal Transport
This work addresses a practical challenge in deploying multimodal AI systems for expression recognition, offering an incremental improvement over existing privileged knowledge distillation methods.
The paper tackles the problem of multimodal expression recognition in real-world settings where some training modalities are unavailable at test time, by proposing a knowledge distillation method using optimal transport to capture structural information from privileged modalities, achieving state-of-the-art performance on pain estimation and arousal-valence prediction tasks.
Deep learning models for multimodal expression recognition have reached remarkable performance in controlled laboratory environments because of their ability to learn complementary and redundant semantic information. However, these models struggle in the wild, mainly because of the unavailability and quality of modalities used for training. In practice, only a subset of the training-time modalities may be available at test time. Learning with privileged information enables models to exploit data from additional modalities that are only available during training. State-of-the-art knowledge distillation (KD) methods have been proposed to distill information from multiple teacher models (each trained on a modality) to a common student model. These privileged KD methods typically utilize point-to-point matching, yet have no explicit mechanism to capture the structural information in the teacher representation space formed by introducing the privileged modality. Experiments were performed on two challenging problems - pain estimation on the Biovid dataset (ordinal classification) and arousal-valance prediction on the Affwild2 dataset (regression). Results show that our proposed method can outperform state-of-the-art privileged KD methods on these problems. The diversity among modalities and fusion architectures indicates that PKDOT is modality- and model-agnostic.