Differentially private transferrable deep learning with membership-mappings

Kumar, Mohit

doi:10.1007/s43674-022-00049-5

Cited by 5 publications

(24 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In contrast, the current study explores geometrically inspired kernel affine hull machines. 2) The input perturbation method (where noise is added to original data to achieve (ǫ, δ)−differential privacy of any subsequent computational algorithm) was earlier considered in Kumar, 2023). However, the current study complements the input perturbation method with a transformation to mitigate the accuracy-loss issue of differential privacy.…”

Section: Kahm Compositions For Data Representation Learning and Class...mentioning

confidence: 99%

“…Algorithm 1 Differentially private approximation of a data matrix (Kumar, 2023) Require: Data matrix Y ∈ R N ×p ; differential privacy parameters:…”

Section: An Optimal (ǫ δ)−Differentially Private Noise Adding Mechanismmentioning

confidence: 99%

“…Recently, a fuzzy theoretic approach to kernel based wide and conditionally deep autoencoders has been introduced Zhang et al, 2022;Zhang et al, 2023;Kumar et al, 2021bKumar et al, , 2021aKumar et al, , 2023, wherein analytical solutions are derived for the learning of models using variational optimization technique. This approach has been further extended to privacy-preserving learning under a differential privacy framework (Kumar, 2023;Kumar, Rossbory, Moser, & Freudenthaler, 2020). As an alternative to the SVM, the idea of affine hull large margin classifier has been investigated (Cevikalp et al, 2010).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

On Mitigating the Utility-Loss in Differentially Private Learning: A New Perspective by a Geometrically Inspired Kernel Approach

Kumar,

Moser,

Fischer

2024

jair

View full text Add to dashboard Cite

Privacy-utility tradeoff remains as one of the fundamental issues of differentially private machine learning. This paper introduces a geometrically inspired kernel-based approach to mitigate the accuracy-loss issue in classification. In this approach, a representation of the affine hull of given data points is learned in Reproducing Kernel Hilbert Spaces (RKHS). This leads to a novel distance measure that hides privacy-sensitive information about individual data points and improves the privacy-utility tradeoff via significantly reducing the risk of membership inference attacks. The effectiveness of the approach is demonstrated through experiments on MNIST dataset, Freiburg groceries dataset, and a real biomedical dataset. It is verified that the approach remains computationally practical. The application of the approach to federated learning is considered and it is observed that the accuracy-loss due to data being distributed is either marginal or not significantly high.

show abstract

Section: Kahm Compositions For Data Representation Learning and Class...mentioning

confidence: 99%

“…Algorithm 1 Differentially private approximation of a data matrix (Kumar, 2023) Require: Data matrix Y ∈ R N ×p ; differential privacy parameters:…”

Section: An Optimal (ǫ δ)−Differentially Private Noise Adding Mechanismmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On Mitigating the Utility-Loss in Differentially Private Learning: A New Perspective by a Geometrically Inspired Kernel Approach

Kumar,

Moser,

Fischer

2024

jair

View full text Add to dashboard Cite

show abstract

“…In this study, the variational membership mappings are leveraged to build the required stochastic inverse models. Membership mappings [14,15] have been introduced as an alternative to deep neural networks in order to address the issues such as determining the optimal model structure, smaller training dataset, and iterative time-consuming nature of numerical learning algorithms [16][17][18][19][20][21][22]. A membership mapping represents data through a fuzzy set (characterized by a membership function such that the dimension of the membership function increases with an increasing data size).…”

Section: Variational Membership Mapping Bayesian Modelsmentioning

confidence: 99%

“…However, in practice, the data distributions are unknown, and thus, a way to approximate the defined measures is required. Therefore, a novel method that employs recently introduced membership mappings [14][15][16][17][18][19][20][21][22], is presented for approximating the defined privacy leakage, interpretability, and transferability measures. The method relies on inferring a variational Bayesian model that facilitates an analytical approximation of the information theoretic measures through variational optimization methodology.…”

Section: Computation Of Information Theoretic Measures Without Requir...mentioning

confidence: 99%

An Information Theoretic Approach to Privacy-Preserving Interpretable and Transferable Learning

Kumar,

Moser,

Fischer

et al. 2023

Algorithms

Self Cite

View full text Add to dashboard Cite

In order to develop machine learning and deep learning models that take into account the guidelines and principles of trustworthy AI, a novel information theoretic approach is introduced in this article. A unified approach to privacy-preserving interpretable and transferable learning is considered for studying and optimizing the trade-offs between the privacy, interpretability, and transferability aspects of trustworthy AI. A variational membership-mapping Bayesian model is used for the analytical approximation of the defined information theoretic measures for privacy leakage, interpretability, and transferability. The approach consists of approximating the information theoretic measures by maximizing a lower-bound using variational optimization. The approach is demonstrated through numerous experiments on benchmark datasets and a real-world biomedical application concerned with the detection of mental stress in individuals using heart rate variability analysis.

show abstract