A method estimating how much removing one training example would have changed a model's parameters and, downstream, its output.
Continue to AI University →