On Second-Order Group Influence Functions for Black-Box Predictions

Samyadeep Basu, Xuchen You, Soheil Feizi
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:715-724, 2020.

Abstract

With the rapid adoption of machine learning systems in sensitive applications, there is an increasing need to make black-box models explainable. Often we want to identify an influential group of training samples in a particular test prediction for a given machine learning model. Existing influence functions tackle this problem by using first-order approximations of the effect of removing a sample from the training set on model parameters. To compute the influence of a group of training samples (rather than an individual point) in model predictions, the change in optimal model parameters after removing that group from the training set can be large. Thus, in such cases, the first-order approximation can be loose. In this paper, we address this issue and propose second-order influence functions for identifying influential groups in test-time predictions. For linear models, across different sizes and types of groups, we show that using the proposed second-order influence function improves the correlation between the computed influence values and the ground truth ones. We also show that second-order influence functions could be used with optimization techniques to improve the selection of the most influential group for a test-sample.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-basu20b, title = {On Second-Order Group Influence Functions for Black-Box Predictions}, author = {Basu, Samyadeep and You, Xuchen and Feizi, Soheil}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {715--724}, year = {2020}, editor = {III, Hal Daumé and Singh, Aarti}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/basu20b/basu20b.pdf}, url = {http://proceedings.mlr.press/v119/basu20b.html}, abstract = {With the rapid adoption of machine learning systems in sensitive applications, there is an increasing need to make black-box models explainable. Often we want to identify an influential group of training samples in a particular test prediction for a given machine learning model. Existing influence functions tackle this problem by using first-order approximations of the effect of removing a sample from the training set on model parameters. To compute the influence of a group of training samples (rather than an individual point) in model predictions, the change in optimal model parameters after removing that group from the training set can be large. Thus, in such cases, the first-order approximation can be loose. In this paper, we address this issue and propose second-order influence functions for identifying influential groups in test-time predictions. For linear models, across different sizes and types of groups, we show that using the proposed second-order influence function improves the correlation between the computed influence values and the ground truth ones. We also show that second-order influence functions could be used with optimization techniques to improve the selection of the most influential group for a test-sample.} }
Endnote
%0 Conference Paper %T On Second-Order Group Influence Functions for Black-Box Predictions %A Samyadeep Basu %A Xuchen You %A Soheil Feizi %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-basu20b %I PMLR %P 715--724 %U http://proceedings.mlr.press/v119/basu20b.html %V 119 %X With the rapid adoption of machine learning systems in sensitive applications, there is an increasing need to make black-box models explainable. Often we want to identify an influential group of training samples in a particular test prediction for a given machine learning model. Existing influence functions tackle this problem by using first-order approximations of the effect of removing a sample from the training set on model parameters. To compute the influence of a group of training samples (rather than an individual point) in model predictions, the change in optimal model parameters after removing that group from the training set can be large. Thus, in such cases, the first-order approximation can be loose. In this paper, we address this issue and propose second-order influence functions for identifying influential groups in test-time predictions. For linear models, across different sizes and types of groups, we show that using the proposed second-order influence function improves the correlation between the computed influence values and the ground truth ones. We also show that second-order influence functions could be used with optimization techniques to improve the selection of the most influential group for a test-sample.
APA
Basu, S., You, X. & Feizi, S.. (2020). On Second-Order Group Influence Functions for Black-Box Predictions. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:715-724 Available from http://proceedings.mlr.press/v119/basu20b.html.

Related Material