Saved in:
Bibliographic Details
Main Authors: Ilyas, Andrew, Engstrom, Logan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.16430
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915254828531712
author Ilyas, Andrew
Engstrom, Logan
author_facet Ilyas, Andrew
Engstrom, Logan
contents The goal of predictive data attribution is to estimate how adding or removing a given set of training datapoints will affect model predictions. In convex settings, this goal is straightforward (i.e., via the infinitesimal jackknife). In large-scale (non-convex) settings, however, existing methods are far less successful -- current methods' estimates often only weakly correlate with ground truth. In this work, we present a new data attribution method (MAGIC) that combines classical methods and recent advances in metadifferentiation to (nearly) optimally estimate the effect of adding or removing training data on model predictions.
format Preprint
id arxiv_https___arxiv_org_abs_2504_16430
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MAGIC: Near-Optimal Data Attribution for Deep Learning
Ilyas, Andrew
Engstrom, Logan
Machine Learning
Computation and Language
Computer Vision and Pattern Recognition
The goal of predictive data attribution is to estimate how adding or removing a given set of training datapoints will affect model predictions. In convex settings, this goal is straightforward (i.e., via the infinitesimal jackknife). In large-scale (non-convex) settings, however, existing methods are far less successful -- current methods' estimates often only weakly correlate with ground truth. In this work, we present a new data attribution method (MAGIC) that combines classical methods and recent advances in metadifferentiation to (nearly) optimally estimate the effect of adding or removing training data on model predictions.
title MAGIC: Near-Optimal Data Attribution for Deep Learning
topic Machine Learning
Computation and Language
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2504.16430