Projects
Explainable feature boosting
Using feature attribution to refine acoustic representations and examine model decisions.
active
The ICMLA 2023 and Applied Intelligence 2024 papers investigate an iterative loop for feature selection in speech emotion recognition. Shapley values guide analysis of feature relevance, redundancy and contribution.
The conference and journal records remain separate, with their own citations and evaluation scope. Their repositories document the corresponding implementations.
Research directions
- 01
Research programs
Interpretable & multimodal learning
Understanding acoustic features and examining how audio, language and visual cues interact in emotion recognition.