Reducing sentiment polarity for demographic attributes in word embeddings using adversarial learning

Sweeney, Chris; Najafian, Maryam

doi:10.1145/3351095.3372837

Cited by 26 publications

(18 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We train the encoder 𝑓 and discriminator 𝑔 using adversarial training so that we hope the embedding learned via encoder 𝑓 can fool the discriminator 𝑔. Training such a network is challenging and we take advantage of adversarial learning since it has shown promising results for other fairness tasks such as removing unfairness in NLP applications [17,40]. We use adversarial learning to de-correlate the relationships between protected status variable 𝑍 and feature vectors encoded via 𝑓 (𝜃 ).…”

Section: Methods 41 Learning Overviewmentioning

confidence: 99%

See 1 more Smart Citation

Towards Fair Deep Anomaly Detection

Zhang

Davidson

2021

Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Cite

Anomaly detection aims to find instances that are considered unusual and is a fundamental problem of data science. Recently, deep anomaly detection methods were shown to achieve superior results particularly in complex data such as images. Our work focuses on deep one-class classification for anomaly detection which learns a mapping only from the normal samples. However, the non-linear transformation performed by deep learning can potentially find patterns associated with social bias. The challenge with adding fairness to deep anomaly detection is to ensure both making fair and correct anomaly predictions simultaneously. In this paper, we propose a new architecture for the fair anomaly detection approach (Deep Fair SVDD) and train it using an adversarial network to de-correlate the relationships between the sensitive attributes and the learned representations. This differs from how fairness is typically added namely as a regularizer or a constraint. Further, we propose two effective fairness measures and empirically demonstrate that existing deep anomaly detection methods are unfair. We show that our proposed approach can remove the unfairness largely with minimal loss on the anomaly detection performance. Lastly, we conduct an in-depth analysis to show the strength and limitations of our proposed model, including parameter analysis, feature visualization, and run-time analysis. CCS CONCEPTS• Computing methodologies → Machine learning algorithms.

show abstract

Section: Methods 41 Learning Overviewmentioning

confidence: 99%

“…[17] shows that demographic information leaks into intermediate representations of neural networks trained on text datasets and applies adversarial learning to mitigate the information leaks. [40] takes the advantages of adversarial networks to reduce word vector sentiment bias for demographic identity terms.…”

Section: Related Workmentioning

confidence: 99%

Towards Fair Deep Anomaly Detection

Zhang

Davidson

2021

Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Cite

show abstract

“…For instance, a model can memorize protected attributes and allow their disclosure later on (Kumar et al 2019). Another model threat regards how language models address human discriminatory biases from their training text corpora Sweeney and Najafian 2020). Moreover, the computation scenario, such as centralized cloud servers or distributed processing architectures, plays an important role in the existence of privacy threats since many of them are related to the misbehavior of components.…”

Section: Privacy Issues For Nlpmentioning

confidence: 99%

“…Model properties, such as vocabulary and neural network layers, can be used by an adversary to perform attacks that end up disclosing private data used for training (Alawad et al 2020). Additional privacy-related issues from NLP models concern biases (Sweeney and Najafian 2020; Gencoglu 2020; Tan and Celis 2019) and the unfair decisions made by biased models (Sweeney and Najafian 2020;Xu et al 2019). Therefore, from the perspective of NLP models, privacy threats are as follows.…”

Section: Threats From Modelsmentioning

confidence: 99%

How to keep text private? A systematic review of deep learning methods for privacy-preserving natural language processing

Sousa¹,

Kern²

2022

Preprint

View full text Add to dashboard Cite

Deep learning (DL) models for natural language processing (NLP) tasks often handle private data, demanding protection against breaches and disclosures. Data protection laws, such as the European Union's General Data Protection Regulation (GDPR), thereby enforce the need for privacy. Although many privacy-preserving NLP methods have been proposed in recent years, no categories to organize them have been introduced yet, making it hard to follow the progress of the literature. To close this gap, this article systematically reviews over sixty DL methods for privacy-preserving NLP published between 2016 and 2020, covering theoretical foundations, privacy-enhancing technologies, and analysis of their suitability for real-world scenarios. First, we introduce a novel taxonomy for classifying the existing methods into three categories: data safeguarding methods, trusted methods, and verification methods. Second, we present an extensive summary of privacy threats, datasets for applications, and metrics for privacy evaluation. Third, throughout the review, we describe privacy issues in the NLP pipeline in a holistic view. Further, we discuss open challenges in privacy-preserving NLP regarding data traceability, computation overhead, dataset size, the prevalence of human biases in embeddings, and the privacy-utility tradeoff. Finally, this review presents future research directions to guide successive research and development of privacy-preserving NLP models.

show abstract

“…Another approach to fairness in toxicity classification focuses on developing methods to mitigate bias in these systems. Several approaches have been proposed to debias word embeddings [29,33], and various methods have been proposed that directly target the outcomes of toxicity classification systems: [35] uses a multi-task learning model with an attention layer that jointly learn demographic information and toxicity in order to reduce bias, whereas [13] takes a causal approach that imposes a fairness penalty on models that predict different scores for comments that differ only in their use of identity terms. To our knowledge, we are the first group to apply tools from Domain Generalization to a fair toxicity classification task.…”

Section: Fairness In Toxic Comment Detectionmentioning

confidence: 99%

Fairness and Robustness in Invariant Learning: A Case Study in Toxicity Classification

Robert¹,

Creager²,

Madras³

et al. 2020

Preprint

View full text Add to dashboard Cite

Robustness is of central importance in machine learning and has given rise to the fields of domain generalization and invariant learning, which are concerned with improving performance on a test distribution distinct from but related to the training distribution. In light of recent work suggesting an intimate connection between fairness and robustness, we investigate whether algorithms from robust ML can be used to improve the fairness of classifiers that are trained on biased data and tested on unbiased data. We apply Invariant Risk Minimization (IRM), a domain generalization algorithm that employs a causal discovery inspired method to find robust predictors, to the task of fairly predicting the toxicity of internet comments. We show that IRM achieves better out-of-distribution accuracy and fairness than Empirical Risk Minimization (ERM) methods, and analyze both the difficulties that arise when applying IRM in practice and the conditions under which IRM will likely be effective in this scenario. We hope that this work will inspire further studies of how robust machine learning methods relate to algorithmic fairness.Preprint. Under review.

show abstract

Reducing sentiment polarity for demographic attributes in word embeddings using adversarial learning

Cited by 26 publications

References 16 publications

Towards Fair Deep Anomaly Detection

Towards Fair Deep Anomaly Detection

How to keep text private? A systematic review of deep learning methods for privacy-preserving natural language processing

Fairness and Robustness in Invariant Learning: A Case Study in Toxicity Classification

Contact Info

Product

Resources

About