PAC-learning

GPTKB entity

Statements (23)
Predicate Object
gptkbp:instanceOf learning theory
gptkbp:assumes examples are drawn independently from a fixed but unknown distribution
gptkbp:describes framework for studying learnability of functions
gptkbp:field gptkb:machine_learning
computational learning theory
gptkbp:fullName gptkb:Probably_Approximately_Correct_learning
gptkbp:goal find a hypothesis that is probably approximately correct
gptkbp:hasConcept algorithm can learn a function with high probability and small error
https://www.w3.org/2000/01/rdf-schema#label PAC-learning
gptkbp:influenced development of modern machine learning theory
gptkbp:introduced gptkb:Leslie_Valiant
gptkbp:introducedIn 1984
gptkbp:mathematicalFoundation gptkb:probability_theory
learning theory
gptkbp:parameter delta (confidence)
epsilon (accuracy)
gptkbp:relatedConcept gptkb:empirical_risk_minimization
gptkb:VC_dimension
sample complexity
agnostic learning
gptkbp:usedIn theoretical analysis of machine learning algorithms
gptkbp:bfsParent gptkb:David_McAllester
gptkbp:bfsLayer 7