Aleatoric uncertainty
Aleatoric uncertainty: Uncertainty caused by intrinsic noise or variability in data, which cannot be removed by collecting more. It includes imperfect measurements, random outcomes and label ambiguity. It is often modeled through the output distribution and distinguished from uncertainty due to limited model knowledge.
In italiano: Aleatoric UncertaintyLegge II · Probabilità, softmax e modelli generativiIt includes imperfect measurements, random outcomes and label ambiguity. It is often modeled through the output distribution and distinguished from uncertainty due to limited model knowledge.
Examples
- Sensor noise
- Two plausible diagnoses
- Trajectory affected by random events