How Many and Which Training Points Would Need to be Removed to Flip this Prediction?

Jinghan Yang, Sarthak Jain, Byron C. Wallace


Abstract
We consider the problem of identifying a minimal subset of training data 𝒮t such that if the instances comprising 𝒮t had been removed prior to training, the categorization of a given test point xt would have been different.Identifying such a set may be of interest for a few reasons.First, the cardinality of 𝒮t provides a measure of robustness (if |𝒮t| is small for xt, we might be less confident in the corresponding prediction), which we show is correlated with but complementary to predicted probabilities.Second, interrogation of 𝒮t may provide a novel mechanism for contesting a particular model prediction: If one can make the case that the points in 𝒮t are wrongly labeled or irrelevant, this may argue for overturning the associated prediction. Identifying 𝒮t via brute-force is intractable.We propose comparatively fast approximation methods to find 𝒮t based on influence functions, and find that—for simple convex text classification models—these approaches can often successfully identify relatively small sets of training examples which, if removed, would flip the prediction.
Anthology ID:
2023.eacl-main.188
Volume:
Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics
Month:
May
Year:
2023
Address:
Dubrovnik, Croatia
Editors:
Andreas Vlachos, Isabelle Augenstein
Venue:
EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2571–2584
Language:
URL:
https://aclanthology.org/2023.eacl-main.188
DOI:
10.18653/v1/2023.eacl-main.188
Bibkey:
Cite (ACL):
Jinghan Yang, Sarthak Jain, and Byron C. Wallace. 2023. How Many and Which Training Points Would Need to be Removed to Flip this Prediction?. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 2571–2584, Dubrovnik, Croatia. Association for Computational Linguistics.
Cite (Informal):
How Many and Which Training Points Would Need to be Removed to Flip this Prediction? (Yang et al., EACL 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.eacl-main.188.pdf
Dataset:
 2023.eacl-main.188.dataset.zip
Video:
 https://aclanthology.org/2023.eacl-main.188.mp4