 Stability (learning theory)

In statistical learning theory, Stability measures how much a learned predictor or its loss changes when the training sample is perturbed. Uniform stability of a learning algorithm imposes a bound valid at every test example.