The Medium Is Not the Message: Deconfounding Document Embeddings via Linear Concept Erasure (2025.emnlp-main)
Copied to clipboard
| Challenge: | Embedding-based similarity metrics can be influenced by content dimensions and spurious attributes like the text’s source or language. |
| Approach: | They propose a debiasing algorithm that removes observed confounders from encoder representations and removes them from the encoder. |
| Outcome: | The proposed method improves on out-of-distribution benchmarks and on benchmarks, but performance is not affected. |
Similar Papers
From Prejudice to Parity: A New Approach to Debiasing Large Language Model Word Embeddings (2025.coling-main)
Copied to clipboard
| Challenge: | Existing work in this field has looked most commonly into gender bias, racial bias, and religious bias. |
| Approach: | They propose an algorithm that uses a neural network to perform ‘soft debiasing’ and build on the seminal work of (CITATION) and (CitATION). |
| Outcome: | The proposed algorithm outperforms current methods on gender, race, and religion metrics on a wide range of metrics. |
Unpacking Bias: An Empirical Study of Bias Measurement Metrics, Mitigation Algorithms, and Their Interactions (2024.lrec-main)
Copied to clipboard
| Challenge: | Word embeddings (WE) models reflect gender, racial, and religious stereotypes from the corpus on which they are trained. |
| Approach: | They propose a method that carefully controls for word sets and vector normalization to address these factors. |
| Outcome: | The proposed method detects consistency between different mitigation methods and the evaluation words used by the mitigation methods. |
A Prompt Array Keeps the Bias Away: Debiasing Vision-Language Models with Adversarial Learning (2022.aacl-main)
Copied to clipboard
| Challenge: | Large-scale, pretrained vision-language models are growing in popularity due to impressive performance on downstream tasks with minimal finetuning. |
| Approach: | They propose to apply ranking metrics to image-text representations to investigate bias measures and debiasing methods to reduce various bias measures. |
| Outcome: | The proposed model reduces bias measures with minimal degradation to image-text representations. |
On the Interpretability and Significance of Bias Metrics in Texts: a PMI-based Approach (2023.acl-short)
Copied to clipboard
| Challenge: | Word embeddings have been used to quantify biases in texts for years, but their statistical properties and advantages have not been studied. |
| Approach: | They propose to use PMI-based metric to quantify bias in corpora by conditional probabilities and odds ratio to approximate it. |
| Outcome: | The proposed measure can be approximated by an odds ratio, which makes statistical inferences cost-effective and meaningful. |
Black is to Criminal as Caucasian is to Police: Detecting and Removing Multiclass Bias in Word Embeddings (N19-1)
Copied to clipboard
| Challenge: | Existing methods to debias word embeddings in binary settings such as gender and religion are limited to binary labels, whereas word2vec embedders can be used to propagate biases. |
| Approach: | They propose a method to debias word embeddings in multiclass settings such as gender and religion, extending the work of Bolukbasi et al. (2016). |
| Outcome: | The proposed method maintains the efficacy in standard NLP tasks while maintaining the utility of embeddings. |
Debiasing knowledge graph embeddings (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to train knowledge graph embeddings to be neutral to sensitive attributes such as gender have been shown to increase training time by a factor of eight or more. |
| Approach: | They propose a method where all embeddings are trained to be neutral to sensitive attributes such as gender by default using an adversarial loss. |
| Outcome: | The proposed method reduces training time by eightfold and improves accuracy. |
Double-Hard Debias: Tailoring Word Embeddings for Gender Bias Mitigation (2020.acl-main)
Copied to clipboard
| Challenge: | Existing methods to debias word embeddings from human-generated corpora inherit strong gender bias . prior work has suggested removing gender component from pre-trained word embeds or compressing gender information into a few dimensions of the embeddable space . |
| Approach: | They propose a technique that purifies word embeddings against inferred gender subspaces . they propose to preserve distributional semantics of pre-trained word embeds while reducing gender bias . |
| Outcome: | The proposed technique preserves distributional semantics of pre-trained word embeddings while reducing gender bias to a larger degree than prior approaches. |
Gender-preserving Debiasing for Pre-trained Word Embeddings (P19-1)
Copied to clipboard
| Challenge: | Existing methods for debiasing word embeddings have shown discriminative biases . word embeds learnt from social media have shown to encode racist, offensive and discriminative language usage. |
| Approach: | They propose a method that preserves gender-related information while removing stereotypical gender biases from pre-trained word embeddings. |
| Outcome: | The proposed method preserves gender-related information while removing stereotypical discriminative gender biases from pre-trained word embeddings. |
What is in a name? Mitigating Name Bias in Text Embedding Similarity via Anonymization (2025.findings-acl)
Copied to clipboard
| Challenge: | Text-embedding models often exhibit name bias due to data on which they are trained. |
| Approach: | They propose a method to mitigate name bias in text-embedding models by removing references to names from the text. |
| Outcome: | The proposed approach achieves significant performance gains on three downstream NLP tasks involving embedding similarities. |
Neutralizing Gender Bias in Word Embeddings with Latent Disentanglement and Counterfactual Generation (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Recent research shows word embeddings have strong gender biases in embeddable spaces . a proposed method can be used to debiase word embeds without loss of semantic information . |
| Approach: | They propose a latent disentanglement method with a siamese auto-encoder structure with an adapted gradient reversal layer to debiase word embeddings. |
| Outcome: | The proposed method can preserve semantic information during debiasing while minimizing loss of semantic information for extrinsic NLP tasks. |