Estimating Training Data Influence by Tracing Gradient Descent

2020 inproceedings pruthi2020tracin ○ Not yet verified

Authors

Garima Pruthi, Frederick Liu, Mukund Sundararajan, Satyen Kale

Venue

Advances in Neural Information Processing Systems (NeurIPS)

Abstract

Introduces TracIn, which computes influence of training examples by tracing how test loss changes during training. Uses first-order gradient approximation and saved checkpoints for scalability.

BibTeX

Local Entry

@inproceedings{pruthi2020tracin,
  title = {Estimating Training Data Influence by Tracing Gradient Descent},
  author = {Garima Pruthi and Frederick Liu and Mukund Sundararajan and Satyen Kale},
  year = {2020},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  url = {https://arxiv.org/abs/2002.08484},
  abstract = {Introduces TracIn, which computes influence of training examples by tracing how test loss changes during training. Uses first-order gradient approximation and saved checkpoints for scalability.}
}

From OPENALEX

@inproceedings{pruthi2020tracin,
  title = {Estimating Training Data Influence by Tracing Gradient Descent},
  author = {Garima Pruthi and Frederick Liu and Mukund Sundararajan and Satyen Kale},
  year = {2020},
  booktitle = {arXiv (Cornell University)},
  doi = {10.48550/arxiv.2002.08484}
}