Concept
Influence function
An influence function estimates how much a single training example changed a trained model, without retraining. Formally, it approximates the effect of upweighting (or removing) one point on the model's parameters or on the loss at some test point, using the gradient of that…
The rest of “Influence function” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→