@inproceedings{wang-liu-2025-beyond,
title = "Beyond Generation: Leveraging {LLM} Creativity to Overcome Label Bias in Classification",
author = "Wang, Xiaoyue and
Liu, Xin",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-acl.1307/",
doi = "10.18653/v1/2025.findings-acl.1307",
pages = "25500--25506",
ISBN = "979-8-89176-256-5",
abstract = "Large Language Models (LLMs) exhibit impressive capabilities in In-Context Learning (ICL) but are prone to label bias{---}an undesirable tendency to favor certain answers. Existing calibration methods mitigate bias by leveraging in-domain data, yet such data is often unavailable in real-world scenarios. To address this limitation, we propose SDC (Synthetic Data Calibration), a simple-yet-effective approach that generates synthetic in-domain data from a few in-context demonstrations and utilizes it for calibration. By approximating the benefits of real in-domain data, SDC effectively reduces label bias without requiring access to actual domain-specific inputs. Experimental evaluations on 279 classification and multiple-choice tasks from the Super-NaturalInstructions benchmark. The results show that SDC significantly reduces label bias, achieving an average Bias Score reduction of 57.5{\%}, and outperforming all competitive baselines. Moreover, when combined with Leave-One-Out Calibration (LOOC), further improves performance, underscoring its effectiveness and generalizability in enhancing the reliability of LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-liu-2025-beyond">
<titleInfo>
<title>Beyond Generation: Leveraging LLM Creativity to Overcome Label Bias in Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xiaoyue</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-256-5</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) exhibit impressive capabilities in In-Context Learning (ICL) but are prone to label bias—an undesirable tendency to favor certain answers. Existing calibration methods mitigate bias by leveraging in-domain data, yet such data is often unavailable in real-world scenarios. To address this limitation, we propose SDC (Synthetic Data Calibration), a simple-yet-effective approach that generates synthetic in-domain data from a few in-context demonstrations and utilizes it for calibration. By approximating the benefits of real in-domain data, SDC effectively reduces label bias without requiring access to actual domain-specific inputs. Experimental evaluations on 279 classification and multiple-choice tasks from the Super-NaturalInstructions benchmark. The results show that SDC significantly reduces label bias, achieving an average Bias Score reduction of 57.5%, and outperforming all competitive baselines. Moreover, when combined with Leave-One-Out Calibration (LOOC), further improves performance, underscoring its effectiveness and generalizability in enhancing the reliability of LLMs.</abstract>
<identifier type="citekey">wang-liu-2025-beyond</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.1307</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.1307/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>25500</start>
<end>25506</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Generation: Leveraging LLM Creativity to Overcome Label Bias in Classification
%A Wang, Xiaoyue
%A Liu, Xin
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F wang-liu-2025-beyond
%X Large Language Models (LLMs) exhibit impressive capabilities in In-Context Learning (ICL) but are prone to label bias—an undesirable tendency to favor certain answers. Existing calibration methods mitigate bias by leveraging in-domain data, yet such data is often unavailable in real-world scenarios. To address this limitation, we propose SDC (Synthetic Data Calibration), a simple-yet-effective approach that generates synthetic in-domain data from a few in-context demonstrations and utilizes it for calibration. By approximating the benefits of real in-domain data, SDC effectively reduces label bias without requiring access to actual domain-specific inputs. Experimental evaluations on 279 classification and multiple-choice tasks from the Super-NaturalInstructions benchmark. The results show that SDC significantly reduces label bias, achieving an average Bias Score reduction of 57.5%, and outperforming all competitive baselines. Moreover, when combined with Leave-One-Out Calibration (LOOC), further improves performance, underscoring its effectiveness and generalizability in enhancing the reliability of LLMs.
%R 10.18653/v1/2025.findings-acl.1307
%U https://aclanthology.org/2025.findings-acl.1307/
%U https://doi.org/10.18653/v1/2025.findings-acl.1307
%P 25500-25506
Markdown (Informal)
[Beyond Generation: Leveraging LLM Creativity to Overcome Label Bias in Classification](https://aclanthology.org/2025.findings-acl.1307/) (Wang & Liu, Findings 2025)
ACL