pystra.active_learning.learning

Use it: User guide · Worked example · Theory

Point selection in independent standard normal coordinates.

Functions

learning_eff	Return expected feasibility, best index (maximum), and stopping flag.
learning_u	Return the U score, best index (minimum), and threshold satisfaction.

Classes

EnsembleLearningFunction	Selection requiring actual replicate predictions, not Gaussian spread.
ExpectedFeasibility	Maximum EFF with a default threshold of 1e-3 in limit-state units.
FBRLearning	Minimum bootstrap classification agreement, abs(B_safe-B_failure)/B.
LearningDecision	Selection from the supplied pool, with a diagnostic threshold flag.
LearningFunction	Stateless selection policy operating on scalar mean/spread predictions.
UFunction	Minimum standardized distance to failure; default threshold is 2.