@inproceedings{jin-etal-2025-unlearning,
title = "Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate",
author = "Jin, Xiaomeng and
Bu, Zhiqi and
Vinzamuri, Bhanukiran and
Ramakrishna, Anil and
Chang, Kai-Wei and
Cevher, Volkan and
Hong, Mingyi",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.563/",
doi = "10.18653/v1/2025.naacl-long.563",
pages = "11278--11294",
ISBN = "979-8-89176-189-6",
abstract = "Machine unlearning has been used to remove unwanted knowledge acquired by large language models (LLMs). In this paper, we examine machine unlearning from an optimization perspective, framing it as a regularized multi-task optimization problem, where one task optimizes a forgetting objective and another optimizes the model performance. In particular, we introduce a normalized gradient difference algorithm, enabling us to have better control over the trade-off between the objectives, while integrating a new, automatic learning rate scheduler. We provide a theoretical analysis and empirically demonstrate the superior performance of among state-of-the-art unlearning methods on the TOFU and MUSE datasets while exhibiting stable training."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jin-etal-2025-unlearning">
<titleInfo>
<title>Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xiaomeng</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiqi</namePart>
<namePart type="family">Bu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bhanukiran</namePart>
<namePart type="family">Vinzamuri</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anil</namePart>
<namePart type="family">Ramakrishna</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai-Wei</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Volkan</namePart>
<namePart type="family">Cevher</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingyi</namePart>
<namePart type="family">Hong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-189-6</identifier>
</relatedItem>
<abstract>Machine unlearning has been used to remove unwanted knowledge acquired by large language models (LLMs). In this paper, we examine machine unlearning from an optimization perspective, framing it as a regularized multi-task optimization problem, where one task optimizes a forgetting objective and another optimizes the model performance. In particular, we introduce a normalized gradient difference algorithm, enabling us to have better control over the trade-off between the objectives, while integrating a new, automatic learning rate scheduler. We provide a theoretical analysis and empirically demonstrate the superior performance of among state-of-the-art unlearning methods on the TOFU and MUSE datasets while exhibiting stable training.</abstract>
<identifier type="citekey">jin-etal-2025-unlearning</identifier>
<identifier type="doi">10.18653/v1/2025.naacl-long.563</identifier>
<location>
<url>https://aclanthology.org/2025.naacl-long.563/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>11278</start>
<end>11294</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate
%A Jin, Xiaomeng
%A Bu, Zhiqi
%A Vinzamuri, Bhanukiran
%A Ramakrishna, Anil
%A Chang, Kai-Wei
%A Cevher, Volkan
%A Hong, Mingyi
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-189-6
%F jin-etal-2025-unlearning
%X Machine unlearning has been used to remove unwanted knowledge acquired by large language models (LLMs). In this paper, we examine machine unlearning from an optimization perspective, framing it as a regularized multi-task optimization problem, where one task optimizes a forgetting objective and another optimizes the model performance. In particular, we introduce a normalized gradient difference algorithm, enabling us to have better control over the trade-off between the objectives, while integrating a new, automatic learning rate scheduler. We provide a theoretical analysis and empirically demonstrate the superior performance of among state-of-the-art unlearning methods on the TOFU and MUSE datasets while exhibiting stable training.
%R 10.18653/v1/2025.naacl-long.563
%U https://aclanthology.org/2025.naacl-long.563/
%U https://doi.org/10.18653/v1/2025.naacl-long.563
%P 11278-11294
Markdown (Informal)
[Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate](https://aclanthology.org/2025.naacl-long.563/) (Jin et al., NAACL 2025)
ACL