New model enhances emotion recognition with uncertainty quantification

Researchers have developed an Evidential Multimodal Transformer with Dempster-Shafer Fusion (EMT-DS) to improve emotion recognition accuracy, particularly when dealing with conflicting sensor data. This new model quantifies uncertainty by distinguishing between lack of evidence and sensor noise, and uses Dempster's Rule of Combination to handle contradictory inputs. EMT-DS demonstrated superior performance across multiple benchmarks, including IEMOCAP, DEAP, KEMDy20, and CMU-MOSEI, while also showing robustness to noise and modality absence. AI

IMPACT Improves robustness and accuracy in multimodal AI systems dealing with uncertain or conflicting data.

RANK_REASON Academic paper detailing a new model and its performance on benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• CMU-MOSEI
• CORE Recommender
• DEAP
• Dempster-Shafer Fusion
• Dempster's Rule of Combination
• Evidential Multimodal Transformer with Dempster-Shafer Fusion
• Hugging Face
• IEMOCAP: interactive emotional dyadic motion capture database
• KEMDy20

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →