Fair Transfer Learning with Missing Protected Attributes

Coston, Amanda; Ramamurthy, Karthikeyan Natesan; Wei, Dennis; Varshney, Kush R.; Speakman, Skyler; Mustahsan, Zairah; Chakraborty, Supriyo

doi:10.1145/3306618.3314236

Cited by 61 publications

(44 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, despite good intentions, not having these relevant attributes, classes or tuples creates obstacles to check the bias constraints, whereas biases on these private-sensitive attributes might still exist due to remaining other attributes correlated with the protected ones. Thus, more work on understanding the interactions between privacy and unfairness [83], and on accurately inferring the missing attributes from the available data is needed [35,42]. This would be part of the checking process of the bias constraints.…”

Section: Formalizing the Tensions Between Privacy And Fairnessmentioning

confidence: 99%

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

2021

View full text Add to dashboard Cite

The increasing use of data-driven decision support systems in industry and governments is accompanied by the discovery of a plethora of bias and unfairness issues in the outputs of these systems. Multiple computer science communities, and especially machine learning, have started to tackle this problem, often developing algorithmic solutions to mitigate biases to obtain fairer outputs. However, one of the core underlying causes for unfairness is bias in training data which is not fully covered by such approaches. Especially, bias in data is not yet a central topic in data engineering and management research. We survey research on bias and unfairness in several computer science domains, distinguishing between data management publications and other domains. This covers the creation of fairness metrics, fairness identification, and mitigation methods, software engineering approaches and biases in crowdsourcing activities. We identify relevant research gaps and show which data management activities could be repurposed to handle biases and which ones might reinforce such biases. In the second part, we argue for a novel data-centered approach overcoming the limitations of current algorithmic-centered methods. This approach focuses on eliciting and enforcing fairness requirements and constraints on data that systems are trained, validated, and used on. We argue for the need to extend database management systems to handle such constraints and mitigation methods. We discuss the associated future research directions regarding algorithms, formalization, modelling, users, and systems.

show abstract

Section: Formalizing the Tensions Between Privacy And Fairnessmentioning

confidence: 99%

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

2021

View full text Add to dashboard Cite

show abstract

“…Ref. [ 7 ] tackles the absence of protected attributes using transfer learning from a different dataset that does have protected attributes. Ref.…”

Section: Related Workmentioning

confidence: 99%

“…The majority of these algorithms assume that the protected attribute is accurately specified for the training dataset, which is then used to mitigate unwanted biases by processing the input dataset or modifying the training algorithm (in-processing) or post-processing the output of the prediction algorithm. However, the protected attribute is often unavailable or anonymized for legal reasons [ 5 , 6 , 7 ].…”

Section: Introductionmentioning

confidence: 99%

Interventional Fairness with Indirect Knowledge of Unobserved Protected Attributes

Galhotra

Shanmugam

Sattigeri

et al. 2021

Entropy

Self Cite

View full text Add to dashboard Cite

The deployment of machine learning (ML) systems in applications with societal impact has motivated the study of fairness for marginalized groups. Often, the protected attribute is absent from the training dataset for legal reasons. However, datasets still contain proxy attributes that capture protected information and can inject unfairness in the ML model. Some deployed systems allow auditors, decision makers, or affected users to report issues or seek recourse by flagging individual samples. In this work, we examine such systems and consider a feedback-based framework where the protected attribute is unavailable and the flagged samples are indirect knowledge. The reported samples are used as guidance to identify the proxy attributes that are causally dependent on the (unknown) protected attribute. We work under the causal interventional fairness paradigm. Without requiring the underlying structural causal model a priori, we propose an approach that performs conditional independence tests on observed data to identify such proxy attributes. We theoretically prove the optimality of our algorithm, bound its complexity, and complement it with an empirical evaluation demonstrating its efficacy on various real-world and synthetic datasets.

show abstract

“…So far, few studies have considered transfer of a model's knowledge on fairness to other domains. In [23], a fair transfer learning problem was addressed via instance-based domain adaptation technique, under the assumption that the sensitive attribute for only one of either the source or the target domain was available. Madras et al [11] partially considered the problem by testing whether the fairness of their method (LAFTR) was preserved over different tasks.…”

Section: B Domain Adaptationmentioning

confidence: 99%

Joint Transfer of Model Knowledge and Fairness Over Domains Using Wasserstein Distance

Yoon

Lee

2020

IEEE Access

View full text Add to dashboard Cite

Owing to the increasing use of machine learning in our daily lives, the problem of fairness has recently become an important topic in machine learning societies. Recent studies regarding fairness in machine learning have been conducted to attempt to ensure statistical independence between individual model predictions and designated sensitive attributes. However, in reality, cases exist in which the sensitive variables of data used for learning models differ from the data upon which the model is applied. In this paper, we investigate a methodology for developing a fair classification model for data with limited or no labels, by transferring knowledge from another data domain where information is fully available. This is done by controlling the Wasserstein distances between relevant distributions. Subsequently, we obtain a fair model that could be successfully applied to two datasets with different sensitive attributes. We present theoretical results validating that our approach provably transfers both classification performance and fairness over domains. Experimental results show that our method does indeed promote fairness for the target domain, while retaining reasonable classification accuracy, and that it often outperforms comparative models in terms of joint fairness.

show abstract

Fair Transfer Learning with Missing Protected Attributes

Cited by 61 publications

References 12 publications

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

Managing bias and unfairness in data for decision support: a survey of machine learning and data engineering approaches to identify and mitigate bias and unfairness within data management and analytics systems

Interventional Fairness with Indirect Knowledge of Unobserved Protected Attributes

Joint Transfer of Model Knowledge and Fairness Over Domains Using Wasserstein Distance

Contact Info

Product

Resources

About