Summary
Uncertainty quantification refers to the estimation of confidence in a prediction by ML models.
Details
Typically two forms of uncertainty exist: epistemic uncertainty, which arises because the model has not seen enough examples to make a firm conclusion and which can thus be improved with more training data; and aleotoric uncertainty, which arises from the intrinsic ambiguity from the input-output mapping, resulting in noise that never goes away no matter how many measurements you take.