Methods · Training Data Attribution
TracIn
Survey sections: §11.2.5 Efficient Influence Estimation Using Training Trajectories: TracIn
TracIn (Garima et al., 2020) estimates training-data influence using gradients evaluated along the training trajectory, without computing the inverse Hessian.
Assumptions and requirements
Its core idea follows from the effect of a single Stochastic Gradient Descent (SGD; Robbins & Monro, 1951) update on the loss of a testing sample.
Intended applications
Unlike Hessian-based influence functions, TracIn does not require matrix inversion.
By examining each checkpoint’s contribution, researchers can identify when a training example had the greatest influence on a downstream prediction.
Limitations
TracInCP approximates the influence of training samples using gradients only at selected checkpoints, which may undermine the estimation accuracy.
Related methods
Used in safety applications
Key papers
Cited in §11.2.5.
- Garima, Frederick Liu, Satyen Kale, and Mukund Sundararajan. Estimating training data influence by tracing gradient descent. In Proceedings of the 34th International Conference on Neural Information Processing Systems, NIPS ’20, Red Hook, NY, USA, 2020. Curran Associates Inc. ISBN 9781713829546.
- Herbert Robbins and Sutton Monro. A Stochastic Approximation Method. The Annals of Mathematical Statistics, 22 (3): 400 – 407, 1951. . URL https://doi.org/10.1214/aoms/1177729586.
Also discussed in: §10.2.1 Trajectory-Based Attribution
Other methods in Training Data Attribution
Verbatim Data Attribution: OLMoTrace · Influence Functions (Gradient-Based Attribution) · Kronecker-Factored Approximate Curvature (K-FAC) · Eigenvalue-Corrected K-FAC (EK-FAC) and ASTRA · TRAK (Tracing with the Randomly-projected After Kernel)