Dynamic Restrained Uncertainty Weighting Loss for Multitask Learning of Vocal Expression
This work addresses multitask learning challenges in vocal expression analysis, which is incremental as it builds on existing weighting methods.
The paper tackled the problem of balancing contributions of multiple tasks in multitask learning for vocal expression recognition, specifically for emotions and demographic traits, and achieved an H-Mean score of 0.394, a substantial improvement over the baseline of 0.335.
We propose a novel Dynamic Restrained Uncertainty Weighting Loss to experimentally handle the problem of balancing the contributions of multiple tasks on the ICML ExVo 2022 Challenge. The multitask aims to recognize expressed emotions and demographic traits from vocal bursts jointly. Our strategy combines the advantages of Uncertainty Weight and Dynamic Weight Average, by extending weights with a restraint term to make the learning process more explainable. We use a lightweight multi-exit CNN architecture to implement our proposed loss approach. The experimental H-Mean score (0.394) shows a substantial improvement over the baseline H-Mean score (0.335).