Fast conformal classification using influence functions

Umang Bhatt, Adrian Weller, Giovanni Cherubin
Proceedings of the Tenth Symposium on Conformal and Probabilistic Prediction and Applications, PMLR 152:303-305, 2021.

Abstract

We use influence functions from robust statistics to speed up full conformal prediction. Traditionally, conformal prediction requires retraining multiple leave-one-out classifiers to calculate p-values for each test point. By using influence functions, we are able to approximate this procedure and to speed up considerably the time complexity.

Cite this Paper

BibTeX
@InProceedings{pmlr-v152-bhatt21a, title = {Fast conformal classification using influence functions}, author = {Bhatt, Umang and Weller, Adrian and Cherubin, Giovanni}, booktitle = {Proceedings of the Tenth Symposium on Conformal and Probabilistic Prediction and Applications}, pages = {303--305}, year = {2021}, editor = {Carlsson, Lars and Luo, Zhiyuan and Cherubin, Giovanni and An Nguyen, Khuong}, volume = {152}, series = {Proceedings of Machine Learning Research}, month = {08--10 Sep}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v152/bhatt21a/bhatt21a.pdf}, url = {https://proceedings.mlr.press/v152/bhatt21a.html}, abstract = {We use influence functions from robust statistics to speed up full conformal prediction. Traditionally, conformal prediction requires retraining multiple leave-one-out classifiers to calculate p-values for each test point. By using influence functions, we are able to approximate this procedure and to speed up considerably the time complexity.} }
Endnote
%0 Conference Paper %T Fast conformal classification using influence functions %A Umang Bhatt %A Adrian Weller %A Giovanni Cherubin %B Proceedings of the Tenth Symposium on Conformal and Probabilistic Prediction and Applications %C Proceedings of Machine Learning Research %D 2021 %E Lars Carlsson %E Zhiyuan Luo %E Giovanni Cherubin %E Khuong An Nguyen %F pmlr-v152-bhatt21a %I PMLR %P 303--305 %U https://proceedings.mlr.press/v152/bhatt21a.html %V 152 %X We use influence functions from robust statistics to speed up full conformal prediction. Traditionally, conformal prediction requires retraining multiple leave-one-out classifiers to calculate p-values for each test point. By using influence functions, we are able to approximate this procedure and to speed up considerably the time complexity.
APA
Bhatt, U., Weller, A. & Cherubin, G.. (2021). Fast conformal classification using influence functions. Proceedings of the Tenth Symposium on Conformal and Probabilistic Prediction and Applications, in Proceedings of Machine Learning Research 152:303-305 Available from https://proceedings.mlr.press/v152/bhatt21a.html.

Related Material