KnowBias: Mitigating Social Bias in LLMs via Know-Bias Neuron Enhancement
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915761246699520 |
|---|---|
| author | Pan, Jinhao Raj, Chahat Mukherjee, Anjishnu Mansouri, Sina Wei, Bowen Yada, Shloka Zhu, Ziwei |
| author_facet | Pan, Jinhao Raj, Chahat Mukherjee, Anjishnu Mansouri, Sina Wei, Bowen Yada, Shloka Zhu, Ziwei |
| contents | Large language models (LLMs) exhibit social biases that reinforce harmful stereotypes, limiting their safe deployment. Most existing debiasing methods adopt a suppressive paradigm by modifying parameters, prompts, or neurons associated with biased behavior; however, such approaches are often brittle, weakly generalizable, data-inefficient, and prone to degrading general capability. We propose \textbf{KnowBias}, a lightweight and conceptually distinct framework that mitigates bias by strengthening, rather than suppressing, neurons encoding bias-knowledge. KnowBias identifies neurons encoding bias knowledge using a small set of bias-knowledge questions via attribution-based analysis, and selectively enhances them at inference time. This design enables strong debiasing while preserving general capabilities, generalizes across bias types and demographics, and is highly data efficient, requiring only a handful of simple yes/no questions and no retraining. Experiments across multiple benchmarks and LLMs demonstrate consistent state-of-the-art debiasing performance with minimal utility degradation. Data and code are available at https://github.com/JP-25/KnowBias. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_21864 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | KnowBias: Mitigating Social Bias in LLMs via Know-Bias Neuron Enhancement Pan, Jinhao Raj, Chahat Mukherjee, Anjishnu Mansouri, Sina Wei, Bowen Yada, Shloka Zhu, Ziwei Artificial Intelligence Large language models (LLMs) exhibit social biases that reinforce harmful stereotypes, limiting their safe deployment. Most existing debiasing methods adopt a suppressive paradigm by modifying parameters, prompts, or neurons associated with biased behavior; however, such approaches are often brittle, weakly generalizable, data-inefficient, and prone to degrading general capability. We propose \textbf{KnowBias}, a lightweight and conceptually distinct framework that mitigates bias by strengthening, rather than suppressing, neurons encoding bias-knowledge. KnowBias identifies neurons encoding bias knowledge using a small set of bias-knowledge questions via attribution-based analysis, and selectively enhances them at inference time. This design enables strong debiasing while preserving general capabilities, generalizes across bias types and demographics, and is highly data efficient, requiring only a handful of simple yes/no questions and no retraining. Experiments across multiple benchmarks and LLMs demonstrate consistent state-of-the-art debiasing performance with minimal utility degradation. Data and code are available at https://github.com/JP-25/KnowBias. |
| title | KnowBias: Mitigating Social Bias in LLMs via Know-Bias Neuron Enhancement |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2601.21864 |