Quantification of Credal Uncertainty: A Distance-Based Approach

1. 1Define credal sets as closed convex sets of probability measures
2. 2Propose a distance-based approach to quantify uncertainty
3. 3Apply the approach to multiclass classification problems
4. 4Evaluate the effectiveness of the approach in capturing total, aleatoric, and epistemic uncertainty

Machine learning researchers and practitioners can benefit from this approach to better understand and quantify uncertainty in their models, particularly in multiclass classification scenarios

💡 Credal sets can be used to represent aleatoric and epistemic uncertainty in machine learning

📊 Quantify uncertainty in ML with a distance-based approach!