Privacy-Preserving Machine Learning

Samaraweera, G. Dumindu; Zhuang, Di; Chang, J. Morris

doi:10.1007/978-981-16-9139-3

Cited by 6 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Trustworthy ML is a vast topic with entire books written about individual pillars [12,19,40,43]. Rather than recounting this entire literature, we provide a brief review of the two pillars with which we have started our effort: fairness and explainability.…”

Section: Background and Related Workmentioning

confidence: 99%

Function Composition in Trustworthy Machine Learning: Implementation Choices, Insights, and Questions

Nagireddy¹,

Singh²,

Hoffman³

et al. 2023

Preprint

View full text Add to dashboard Cite

Ensuring trustworthiness in machine learning (ML) models is a multi-dimensional task. In addition to the traditional notion of predictive performance, other notions such as privacy, fairness, robustness to distribution shift, adversarial robustness, interpretability, explainability, and uncertainty quantification are important considerations to evaluate and improve (if deficient). However, these sub-disciplines or 'pillars' of trustworthiness have largely developed independently, which has limited us from understanding their interactions in real-world ML pipelines. In this paper, focusing specifically on compositions of functions arising from the different pillars, we aim to reduce this gap, develop new insights for trustworthy ML, and answer questions such as the following. Does the composition of multiple fairness interventions result in a fairer model compared to a single intervention? How do bias mitigation algorithms for fairness affect local post-hoc explanations? Does a defense algorithm for untargeted adversarial attacks continue to be effective when composed with a privacy transformation? Toward this end, we report initial empirical results and new insights from 9 different compositions of functions (or pipelines) on 7 real-world datasets along two trustworthy dimensions -fairness and explainability. We also report progress, and implementation choices, on an extensible composer tool to encourage the combination of functionalities from multiple pillars. To-date, the tool supports bias mitigation algorithms for fairness and post-hoc explainability methods. We hope this line of work encourages the thoughtful consideration of multiple pillars when attempting to formulate and resolve a trustworthiness problem.

show abstract

Section: Background and Related Workmentioning

confidence: 99%

Function Composition in Trustworthy Machine Learning: Implementation Choices, Insights, and Questions

Nagireddy¹,

Singh²,

Hoffman³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Even if data and ML models remain inaccessible and only test results are revealed, it is still possible to gain insights into private datasets. Model inversion and membership inference attacks can also target trained ML models [9], [10]. Consequently, numerous emerging applications demand efficient and auditable privacy-preserving ML solutions.…”

Section: Introductionmentioning

confidence: 99%

Efficient Privacy-Preserving Framework for Federated and Blockchain Networks

Khordadpour¹

2023

Preprint

View full text Add to dashboard Cite

<p>Machine learning (ML) models, at the core of Artificial intelligence (AI), are widely applied in our digital world. To build these models, huge amounts of data must be collected, and many assets must be protected under privacy law. Data privacy is a critical issue when training and testing ML models. For privacy concerns to be adequately addressed in today’s ML systems, there needs to be considered privacy gaps in ML, as trained ML models can be vulnerable to adversarial attacks. In this regard, federated learning and blockchain networks are the new paradigms that have emerged with the promise of privacy-preserving by design while utilizing ML models. The new paradigms have promising privacy-preserving potential; however, they neglect several fundamental privacy and security issues the fact that adversaries can exploit shared gradients and global parameters, and the parameter server may drop gradients that have been mistakenly or deliberately updated. Also, enough data is available to train models. The proposed research addresses privacy concerns in federated and distributed environments. This is a general overview to establish a privacy-preserving framework that can provide privacy in federated and blockchain-based networks while address to fundamental privacy and security issues.</p>

show abstract

Trust-based secure federated learning framework to mitigate internal attacks for intelligent vehicular networks

Naik,

Dondeti

2025

Peer-to-Peer Netw. Appl.

View full text Add to dashboard Cite

Privacy-Preserving Machine Learning

Cited by 6 publications

References 0 publications

Function Composition in Trustworthy Machine Learning: Implementation Choices, Insights, and Questions

Function Composition in Trustworthy Machine Learning: Implementation Choices, Insights, and Questions

Efficient Privacy-Preserving Framework for Federated and Blockchain Networks

Trust-based secure federated learning framework to mitigate internal attacks for intelligent vehicular networks

Contact Info

Product

Resources

About