Epistemic uncertainty
Epistemic uncertainty is uncertainty in a model’s predictions arising from limited knowledge about the system or data being modeled.
Grokking Bayes
Epistemic uncertainty is reducible uncertainty caused by incomplete knowledge, such as insufficient data about the underlying function.
- How it is used
- Frames overconfident predictions in data-sparse or extrapolation regions and motivates Bayesian neural networks.
- How deep it goes
- Explains the definition, contrasts aleatoric noise, and connects uncertainty to Bayesian weights without deriving inference algorithms.
- Examples
- Uses noisy sine data, neural-network extrapolation, and multiple plausible linear or neural-network models.
- What you can do after
- Evaluate/understand why single-valued predictions can conceal uncertainty and how Bayesian models expose it.
Machine Learning System Design
Epistemic uncertainty is uncertainty caused by limited knowledge or insufficient coverage of the input space in the training data.
- How it is used
- Guides decisions about whether additional training data can improve model performance.
- How deep it goes
- Defines the cause and contrasts it with aleatoric uncertainty, then connects it to data-collection decisions without deriving uncertainty measures.
- Examples
- Uses thinly covered or out-of-scope inputs and a schematic comparison of epistemic and aleatoric uncertainty.
- What you can do after
- Evaluate whether collecting more training data is likely to improve performance and justify its economic cost.
Read how Machine Learning System Design explains epistemic uncertainty →
Also covered in
What to know first
- neural networkA neural network is a computational model composed of interconnected, parameterized units or transformations that learn to map inputs to outputs.
- decision boundaryA decision boundary is a boundary in input or feature space separating regions assigned different classes or decision outcomes by a predictive model.
Related concepts
- uncertaintykind of · Uncertainty is incomplete knowledge or limited confidence about a value, state, interpretation, behavior, or outcome.
- aleatoric uncertaintycontrasts with · Aleatoric uncertainty is uncertainty attributable to inherent randomness or variability in observed data or outcomes.
- Monte Carlo dropoutproduced by · Monte Carlo dropout is an inference technique that retains stochastic dropout across repeated forward passes to approximate a distribution of model predictions.
Questions about epistemic uncertainty
What is epistemic uncertainty?
Epistemic uncertainty is uncertainty in a model’s predictions arising from limited knowledge about the system or data being modeled.
Which Manning books explain epistemic uncertainty?
Grokking Bayes and Machine Learning System Design.
What should I know before learning epistemic uncertainty?
Manning books that teach epistemic uncertainty usually introduce neural network and decision boundary first.
What is the difference between epistemic uncertainty and aleatoric uncertainty?
Epistemic uncertainty is uncertainty in a model’s predictions arising from limited knowledge about the system or data being modeled. Aleatoric uncertainty is uncertainty attributable to inherent randomness or variability in observed data or outcomes.
What is epistemic uncertainty a kind of?
Epistemic 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.

