# Modular and On-demand Bias Mitigation with Attribute-Removal Subnetworks

arXiv: https://arxiv.org/abs/2205.15171 · HTML: https://mlsanity.com/paper/2205.15171/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- Builds on 47 of its 60 references that are in the map.
- First posted on arXiv 30 May 2022.

## Builds on (most cited first, 30 of 47)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [Parameter-Efficient Transfer Learning for NLP](https://mlsanity.com/paper/1902.00751/) (2019)
- [Domain-Adversarial Training of Neural Networks](https://mlsanity.com/paper/1505.07818/) (2015)
- [The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks](https://mlsanity.com/paper/1803.03635/) (2018)
- [Unsupervised Domain Adaptation by Backpropagation](https://mlsanity.com/paper/1409.7495/) (2014)
- [Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings](https://mlsanity.com/paper/1607.06520/) (2016)
- [BitFit: Simple Parameter-efficient Fine-tuning for Transformer-based Masked Language-models](https://mlsanity.com/paper/2106.10199/) (2021)
- [Language (Technology) is Power: A Critical Survey of "Bias" in NLP](https://mlsanity.com/paper/2005.14050/) (2020)
- [AdapterFusion: Non-Destructive Task Composition for Transfer Learning](https://mlsanity.com/paper/2005.00247/) (2020)
- [Deep Domain Confusion: Maximizing for Domain Invariance](https://mlsanity.com/paper/1412.3474/) (2014)
- [Learning Sparse Neural Networks through L0 Regularization](https://mlsanity.com/paper/1712.01312/) (2017)
- [Learning multiple visual domains with residual adapters](https://mlsanity.com/paper/1705.08045/) (2017)
- [The Woman Worked as a Babysitter: On Biases in Language Generation](https://mlsanity.com/paper/1909.01326/) (2019)
- [Compacter: Efficient Low-Rank Hypercomplex Adapter Layers](https://mlsanity.com/paper/2106.04647/) (2021)
- [Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection](https://mlsanity.com/paper/2004.07667/) (2020)
- [Learning Adversarially Fair and Transferable Representations](https://mlsanity.com/paper/1802.06309/) (2018)
- [Parameter-Efficient Transfer Learning with Diff Pruning](https://mlsanity.com/paper/2012.07463/) (2020)
- [Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting](https://mlsanity.com/paper/1901.09451/) (2019)
- [Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP](https://mlsanity.com/paper/2103.00453/) (2021)
- [The Lottery Ticket Hypothesis for Pre-trained BERT Networks](https://mlsanity.com/paper/2007.12223/) (2020)
- [Large Scale Crowdsourcing and Characterization of Twitter Abusive Behavior](https://mlsanity.com/paper/1802.00393/) (2018)
- [Evaluating Gender Bias in Machine Translation](https://mlsanity.com/paper/1906.00591/) (2019)
- [Supermasks in Superposition](https://mlsanity.com/paper/2006.14769/) (2020)
- [AdapterDrop: On the Efficiency of Adapters in Transformers](https://mlsanity.com/paper/2010.11918/) (2020)
- [Training Neural Networks with Fixed Sparse Masks](https://mlsanity.com/paper/2111.09839/) (2021)
- [Deconstructing Lottery Tickets: Zeros, Signs, and the Supermask](https://mlsanity.com/paper/1905.01067/) (2019)
- [Gender Bias in Contextualized Word Embeddings](https://mlsanity.com/paper/1904.03310/) (2019)
- [Demographic Dialectal Variation in Social Media: A Case Study of African-American English](https://mlsanity.com/paper/1608.08868/) (2016)
- [Block Pruning For Faster Transformers](https://mlsanity.com/paper/2109.04838/) (2021)

## Built on by (most cited first, 18 of 18)

- [Bias and Fairness in Large Language Models: A Survey](https://mlsanity.com/paper/2309.00770/) (2023)
- [A Survey on Fairness in Large Language Models](https://mlsanity.com/paper/2308.10149/) (2023)
- [Deconstructing The Ethics of Large Language Models from Long-standing Issues to New-emerging Dilemmas: A Survey](https://mlsanity.com/paper/2406.05392/) (2024)
- [Tackling Bias in Pre-trained Language Models: Current Trends and Under-represented Societies](https://mlsanity.com/paper/2312.01509/) (2023)
- [Natural Language Processing for Human Resources: A Survey](https://mlsanity.com/paper/2410.16498/) (2024)
- [Effective Controllable Bias Mitigation for Classification and Retrieval using Gate Adapters](https://mlsanity.com/paper/2401.16457/) (2024)
- [Hate Speech Detection with Generalizable Target-aware Fairness](https://mlsanity.com/paper/2406.00046/) (2024)
- [More of the Same: Persistent Representational Harms Under Increased Representation](https://mlsanity.com/paper/2503.00333/) (2025)
- [Attention Pruning: Automated Fairness Repair of Language Models via Surrogate Simulated Annealing](https://mlsanity.com/paper/2503.15815/) (2025)
- [Unlabeled Debiasing in Downstream Tasks via Class-wise Low Variance Regularization](https://mlsanity.com/paper/2409.19541/) (2024)
- [Local Contrastive Editing of Gender Stereotypes](https://mlsanity.com/paper/2410.17739/) (2024)
- [Simultaneous Unlearning of Multiple Protected User Attributes From Variational Autoencoder Recommenders Using Adversarial Training](https://mlsanity.com/paper/2410.20965/) (2024)
- [Mitigating Confounding in Speech-Based Dementia Detection through Weight Masking](https://mlsanity.com/paper/2506.05610/) (2025)
- [Leveraging Domain Knowledge for Inclusive and Bias-aware Humanitarian Response Entry Classification](https://mlsanity.com/paper/2305.16756/) (2023)
- [ScaLearn: Simple and Highly Parameter-Efficient Task Transfer by Learning to Scale](https://mlsanity.com/paper/2310.01217/) (2023)
- [PEFTDebias : Capturing debiasing information using PEFTs](https://mlsanity.com/paper/2312.00434/) (2023)
- [FairLangProc: A Python package for fairness in NLP](https://mlsanity.com/paper/2508.03677/) (2025)
- [FairNVT: Fair Classification via Noise Injection in Vision Transformers](https://mlsanity.com/paper/2604.16780/) (2026)

## Abstract (arXiv)

Societal biases are reflected in large pre-trained language models and their fine-tuned versions on downstream tasks. Common in-processing bias mitigation approaches, such as adversarial training and mutual information removal, introduce additional optimization criteria, and update the model to reach a new debiased state. However, in practice, end-users and practitioners might prefer to switch back to the original model, or apply debiasing only on a specific subset of protected attributes. To enable this, we propose a novel modular bias mitigation approach, consisting of stand-alone highly sparse debiasing subnetworks, where each debiasing module can be integrated into the core model on-demand at inference time. Our approach draws from the concept of diff pruning, and proposes a novel training regime adaptable to various representation disentanglement optimizations. We conduct experiments on three classification tasks with gender, race, and age as protected attributes. The results show that our modular approach, while maintaining task performance, improves (or at least remains on-par with) the effectiveness of bias mitigation in comparison with baseline finetuning. Particularly on a two-attribute dataset, our approach with separately learned debiasing subnetworks shows effective utilization of either or both the subnetworks for selective bias mitigation.