Achieving Differential Privacy and Fairness in Logistic Regression

Xu, Danfeng; Yuan, Shuhan; Wu, Xintao

doi:10.1145/3308560.3317584

Cited by 52 publications

(58 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Also, computer science research suggests that including specifically sensitive data such as gender, age, race, and religion would help create fair a gorithms (nondiscriminatory on the grounds of these data) (Galdon Clavell et al 2020), which is also in conflict with protecting and respecting the privacy of consumers. How to meet both privacy and fairness requirements simultaneously in algorithms is an emerging field in computer science (Xu, Yuan, and Wu 2019). However, this question also needs to be addressed by advertising scholars.…”

Section: Tension Between the Advertising Industry And Consumer Advocamentioning

confidence: 99%

Macro and Exogenous Factors in Computational Advertising: Key Issues and New Research Directions

Helberger

Huh

Milne

et al. 2020

Journal of Advertising

View full text Add to dashboard Cite

To advance the emerging research field of computational advertising this article describes the new computational advertising ecosystem, identifies key actors within it and interactions among them, and discusses future research agendas. Specifically, we propose systematic conceptualization for the redefined advertising industry, consumers, government, and technology environmental factors, and discuss emerging and anticipated tensions that arise in the macro and exogenous factors surrounding the new computational advertising industry, leading to suggestions for future research directions. From multidisciplinary angles, areas of tension and related research questions are explored from advertising, business, computer science, and legal perspectives. The proposed research agendas include exploring transparency of computational advertising practice and consumer education; understanding the trade-off between explainability and performance of algorithms; exploring the issue of new consumers as free data laborers, data as commodity, and related consumer agency challenges; understanding the relationship between algorithmic transparency and consumers' literacy; evaluating the trade-off between algorithmic fairness and privacy protection; examining legal and regulatory issues regarding power imbalance between actors in the computational advertising ecosystem; and studying the trade-off between technological innovation and consumer protection and empowerment.

show abstract

Section: Tension Between the Advertising Industry And Consumer Advocamentioning

confidence: 99%

Macro and Exogenous Factors in Computational Advertising: Key Issues and New Research Directions

Helberger

Huh

Milne

et al. 2020

Journal of Advertising

View full text Add to dashboard Cite

show abstract

“…Recently, researchers have attempted to adopt differential privacy to simultaneously achieve both fairness and privacy preservation [52], [53]. This research is motivated by settings where models are required to be non-discriminatory in terms of certain attributes, but these attributes may be sensitive and so must be protected while training the model [54].…”

Section: Applying Differential Privacy To Improve Fairnessmentioning

confidence: 99%

More Than Privacy: Applying Differential Privacy in Key Areas of Artificial Intelligence

Zhu¹,

Ye²,

Wang³

et al. 2022

IEEE Trans. Knowl. Data Eng.

View full text Add to dashboard Cite

Artificial Intelligence (AI) has attracted a great deal of attention in recent years. However, alongside all its advancements, problems have also emerged, such as privacy violations, security issues and model fairness. Differential privacy, as a promising mathematical model, has several attractive properties that can help solve these problems, making it quite a valuable tool. For this reason, differential privacy has been broadly applied in AI but to date, no study has documented which differential privacy mechanisms can or have been leveraged to overcome its issues or the properties that make this possible. In this paper, we show that differential privacy can do more than just privacy preservation. It can also be used to improve security, stabilize learning, build fair models, and impose composition in selected areas of AI. With a focus on regular machine learning, distributed machine learning, deep learning, and multi-agent systems, the purpose of this article is to deliver a new view on many possibilities for improving AI performance with differential privacy techniques.

show abstract

“…Many works [33][34][35][36][37][38][39][40][41] have tried to address fairness and privacy guarantees together. Kilbertus et al [34] is one of the first proposals that addressed the need for combining fairness requirements with privacy guarantees.…”

Section: Related Workmentioning

confidence: 99%

“…They provide a different approach based on Differential Privacy, where privacy is guaranteed through an injectable amount of noise able to mask the presence of a protected individual in a particular dataset. In this direction, other works [36][37][38][39] aimed to learn fair and differentially private ML models. Cummings et al [36], while showing that it is impossible to achieve both differential privacy and exact fairness without non-trivial accuracy, provides a Probably Approximately Correct [107] learner that is differentially private and approximately (with high probability) fair.…”

Section: Related Workmentioning

confidence: 99%

“…Unfortunately few works in the literature have tried to address more than one of these problems simultaneously. Some have tried to face two of them: for example some works combine fairness with privacy [33][34][35][36][37][38][39][40][41], others [42][43][44] combine adversarial learning with fairness, fairness with explainability [45][46][47], adversarial learning with explainability [48,49], and adversarial learning with privacy [50][51][52][53]. For this reason, in this work, we drive toward the development of systems able to ensure trustworthiness by delivering privacy, fairness, and explainability by design.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Toward Learning Trustworthily from Data Combining Privacy, Fairness, and Explainability: An Application to Face Recognition

Franco

Oneto

Navarin

et al. 2021

Entropy

View full text Add to dashboard Cite

In many decision-making scenarios, ranging from recreational activities to healthcare and policing, the use of artificial intelligence coupled with the ability to learn from historical data is becoming ubiquitous. This widespread adoption of automated systems is accompanied by the increasing concerns regarding their ethical implications. Fundamental rights, such as the ones that require the preservation of privacy, do not discriminate based on sensible attributes (e.g., gender, ethnicity, political/sexual orientation), or require one to provide an explanation for a decision, are daily undermined by the use of increasingly complex and less understandable yet more accurate learning algorithms. For this purpose, in this work, we work toward the development of systems able to ensure trustworthiness by delivering privacy, fairness, and explainability by design. In particular, we show that it is possible to simultaneously learn from data while preserving the privacy of the individuals thanks to the use of Homomorphic Encryption, ensuring fairness by learning a fair representation from the data, and ensuring explainable decisions with local and global explanations without compromising the accuracy of the final models. We test our approach on a widespread but still controversial application, namely face recognition, using the recent FairFace dataset to prove the validity of our approach.

show abstract

Achieving Differential Privacy and Fairness in Logistic Regression

Cited by 52 publications

References 14 publications

Macro and Exogenous Factors in Computational Advertising: Key Issues and New Research Directions

Macro and Exogenous Factors in Computational Advertising: Key Issues and New Research Directions

More Than Privacy: Applying Differential Privacy in Key Areas of Artificial Intelligence

Toward Learning Trustworthily from Data Combining Privacy, Fairness, and Explainability: An Application to Face Recognition

Contact Info

Product

Resources

About