Aleatoric uncertainty
Aleatoric uncertainty is uncertainty attributable to inherent randomness or variability in observed data or outcomes.
Grokking Bayes
Aleatoric uncertainty is irreducible uncertainty caused by inherent randomness or noise in the data-generating process.
- How it is used
- Frames predictive uncertainty as a requirement for interpreting noisy observations and model predictions.
- How deep it goes
- Defines irreducible data noise and contrasts it with epistemic uncertainty, stopping short of a detailed noise model.
- Examples
- Uses noisy sine-function observations, standard neural-network predictions, and extrapolation beyond sparse training data.
- What you can do after
- Evaluate predictive outputs by separating irreducible observation noise from uncertainty caused by limited model knowledge.
Machine Learning System Design
Aleatoric uncertainty is irreducible uncertainty arising from randomness, noise, or ambiguity inherent in the data-generating process.
- How it is used
- Uses aleatoric uncertainty to decide whether an ill-posed ML problem is worth tackling and whether additional data can help.
- How deep it goes
- Defines irreducible data uncertainty, contrasts it with epistemic uncertainty, and stops short of methods for modeling it.
- Examples
- Uses casino roulette, overlapping classes, and additive noise to illustrate irreducibility.
- What you can do after
- Evaluate whether collecting more data can reduce uncertainty or whether the problem is intrinsically noisy.
Read how Machine Learning System Design explains aleatoric uncertainty →
Also covered in
Related concepts
- epistemic uncertaintycontrasts with · Epistemic uncertainty is uncertainty in a model’s predictions arising from limited knowledge about the system or data being modeled.
- uncertaintykind of · Uncertainty is incomplete knowledge or limited confidence about a value, state, interpretation, behavior, or outcome.
- Monte Carlo dropoutuses · Monte Carlo dropout is an inference technique that retains stochastic dropout across repeated forward passes to approximate a distribution of model predictions.
- datasetcontrasts with · A dataset is an organized collection of related data items, records, observations, or examples treated as a unit for storage, processing, or analysis.
Questions about aleatoric uncertainty
What is aleatoric uncertainty?
Aleatoric uncertainty is uncertainty attributable to inherent randomness or variability in observed data or outcomes.
Which Manning books explain aleatoric uncertainty?
Grokking Bayes and Machine Learning System Design.
What is the difference between aleatoric uncertainty and epistemic uncertainty?
Aleatoric uncertainty is uncertainty attributable to inherent randomness or variability in observed data or outcomes. Epistemic uncertainty is uncertainty in a model’s predictions arising from limited knowledge about the system or data being modeled.
What is aleatoric uncertainty a kind of?
Aleatoric uncertainty is a kind of uncertainty. Uncertainty is incomplete knowledge or limited confidence about a value, state, interpretation, behavior, or outcome.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.

