Iterative Distillation for Better Uncertainty Estimates in Multitask Emotion Recognition
This work addresses emotion recognition uncertainty for applications like human-computer interaction, but it is incremental as it builds on existing distillation and ensemble techniques.
The paper tackled the problem of modeling emotion uncertainty from single emotion labels, proposing iterative self-distillation with deep ensembles to improve uncertainty estimates. The method outperformed Temperature Scaling and Monte Carlo Dropout on the Aff-wild2 dataset, providing more reliable uncertainty estimates for in-domain and out-of-domain samples.
When recognizing emotions, subtle nuances in displays of emotion generate ambiguity or uncertainty in emotion perception. Emotion uncertainty has been previously interpreted as inter-rater disagreement among multiple annotators. In this paper, we consider a more common and challenging scenario: modeling emotion uncertainty when only single emotion labels are available. From a Bayesian perspective, we propose to use deep ensembles to capture uncertainty for multiple emotion descriptors, i.e., action units, discrete expression labels and continuous descriptors. We further apply iterative self-distillation. Iterative distillation over multiple generations significantly improves performance in both emotion recognition and uncertainty estimation. Our method generates single student models that provide accurate estimates of uncertainty for in-domain samples and a student ensemble that can detect out-of-domain samples. Our experiments on emotion recognition and uncertainty estimation using the Aff-wild2 dataset demonstrate that our algorithm gives more reliable uncertainty estimates than both Temperature Scaling and Monte Carol Dropout.