1 paper
Jin Hwa Lee, Matthew Smith, Maxwell Adam +1
Current training data attribution (TDA) methods treat the influence one sample has on another as static, but neural networks learn in distinct stages that exhibit changing patterns…