Bayes optimality

E766787

Bayes optimality is a criterion in statistical decision theory under which a decision rule minimizes expected loss with respect to a given prior distribution, making it the benchmark for comparing and justifying optimal procedures.

Try in SPARQL Jump to: Statements Referenced by

Statements (45)

Predicate Object
instanceOf concept in Bayesian statistics
concept in statistical decision theory
decision-theoretic criterion
appliesTo classification
hypothesis testing
point estimation
prediction
assumes specified prior over unknown parameters
basedOn Bayes risk minimization NERFINISHED
benchmarkFor comparing statistical decision rules
justifying optimal procedures
characterizedBy dependence on a specified prior distribution
minimization of posterior expected loss
contrastedWith admissibility
minimax optimality
criterionFor decision rule
statistical procedure
definedAs property of a decision rule that minimizes expected loss with respect to a prior distribution
dependsOn choice of loss function
choice of prior distribution
evaluatedBy Bayes risk functional
field Bayesian statistics
statistical decision theory
formalizedIn modern decision theory
goal minimize Bayes risk
historicalRoot work of Thomas Bayes
implies no other decision rule has lower expected loss under the given prior
influences design of Bayesian classifiers
design of Bayesian estimators
invariantUnder equivalent reparameterizations of the model (given transformed prior and loss)
mathematicalNature optimization problem over decision rules
relatedTo Bayes classifier NERFINISHED
Bayes estimator
Bayes rule NERFINISHED
requires probabilistic model for data
specification of action space
typicalAssumption rational decision maker minimizing expected loss
usedIn econometrics
machine learning
pattern recognition
signal processing
usesConcept Bayes risk NERFINISHED
expected loss
loss function
prior distribution

Referenced by (1)

Full triples — surface form annotated when it differs from this entity's canonical label.