Shared References

← Back to browse

Estimating Training Data Influence by Tracing Gradient Descent

2020 inproceedings pruthi2020tracin
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.