Uncertainty quantification with classification models for better explainability and human-in-the-Loop systems.
Nancy
IA
Uncertainty quantification is a broad field that extends beyond artificial intelligence. Lying at the intersection of mathematics and computer science, its goal is to model and quantify the uncertainty of a given source. In machine learning, the uncertainty of the model itself is crucial for various tasks such as prediction, decision-making, and incremental or active learning. A model’s hesitation can be informative, and many techniques can be applied to model this uncertainty in different forms. Understanding and explaining a model’s prediction is one of the major challenges in artificial intelligence.
Back to top