In this section
@inproceedings{Zylberajch:2021:v1/2021.internlp-1.1,
author = {Zylberajch, H and Lertvittayakumjorn, P and Toni, F},
doi = {v1/2021.internlp-1.1},
pages = {1--6},
publisher = {ASSOC COMPUTATIONAL LINGUISTICS-ACL},
title = {HILDIF: interactive debugging of NLI models using influence functions},
url = {http://dx.doi.org/10.18653/v1/2021.internlp-1.1},
year = {2021}
}
TY - CPAPER
AB - Biases and artifacts in training data can cause unwelcome behavior in text classifiers (such as shallow pattern matching), leading to lack of generalizability. One solution to this problem is to include users in the loop and leverage their feedback to improve models. We propose a novel explanatory debugging pipeline called HILDIF, enabling humans to improve deep text classifiers using influence functions as an explanation method. We experiment on the Natural Language Inference (NLI) task, showing that HILDIF can effectively alleviate artifact problems in fine-tuned BERT models and result in increased model generalizability.
AU - Zylberajch,H
AU - Lertvittayakumjorn,P
AU - Toni,F
DO - v1/2021.internlp-1.1
EP - 6
PB - ASSOC COMPUTATIONAL LINGUISTICS-ACL
PY - 2021///
SP - 1
TI - HILDIF: interactive debugging of NLI models using influence functions
UR - http://dx.doi.org/10.18653/v1/2021.internlp-1.1
UR - http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=LinksAMR&KeyUT=WOS:000696714000001&DestLinkType=FullRecord&DestApp=ALL_WOS&UsrCustomerID=1ba7043ffcc86c417c072aa74d649202
UR - https://aclanthology.org/2021.internlp-1.1/
UR - http://hdl.handle.net/10044/1/93892
ER -