Federated Principal Component Analysis for Genome-Wide Association Studies

Hartebrodt, Anne; Nasirigerdeh, Reza; Blumenthal, David B.; Röttger, Richard

doi:10.1109/icdm51629.2021.00127

Cited by 14 publications

(11 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the federated domain, SVD has been studied extensively, and multiple algorithms exist (e. g. [4], [7], [11]). Given the vertically distributed matrix A s ∈ R m×n s with dimension m × n s at sites s the federated singular value decomposition is defined as…”

Section: Centralized Singular Value Decompositionmentioning

confidence: 99%

“…where U is the full left singular vector and V s are the partial right singular vectors. The right singular vectors should not be shared due to potential privacy breaches [4].…”

Section: Centralized Singular Value Decompositionmentioning

confidence: 99%

“…We recently identified federated QR orthonormalisation as a contributor to a more privacy preserving principal component analysis (PCA) algorithm [4]. Using federated QR orthonormalization for singular value decomposition allows the right, patient-associated singular vectors to remain private when using federated power iteration.…”

Section: Introductionmentioning

confidence: 99%

“…The primary goal of this article is to evaluate the data disclosure of the three algorithms when deployed in a federated setting. In an earlier article, we introduced federated Gram-Schmidt orthonormalisation [4], but it does not return the full decomposition. Therefore, it needs to be extended to return a full QR factorization.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Privacy of federated QR decomposition using additive secure multiparty computation

Hartebrodt¹,

Röttger²

2022

Preprint

View full text Add to dashboard Cite

Federated learning (FL) is a privacy-aware data mining strategy keeping the private data on the owners' machine and thereby confidential. The clients compute local models and send them to an aggregator which computes a global model. In hybrid FL, the local parameters are additionally masked using secure aggregation, such that only the global aggregated statistics become available in clear text, not the client specific updates. Federated QR decomposition has not been studied extensively in the context of cross-silo federated learning. In this article, we investigate the suitability of three QR decomposition algorithms for cross-silo FL and suggest a privacy-aware QR decomposition scheme based on the Gram-Schmidt algorithm which does not blatantly leak raw data. We apply the algorithm to compute linear regression in a federated manner.

show abstract

Section: Centralized Singular Value Decompositionmentioning

confidence: 99%

“…where U is the full left singular vector and V s are the partial right singular vectors. The right singular vectors should not be shared due to potential privacy breaches [4].…”

Section: Centralized Singular Value Decompositionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Privacy of federated QR decomposition using additive secure multiparty computation

Hartebrodt¹,

Röttger²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…• This algorithm is generically applicable for federated singular value decomposition on both "horizontally" and "vertically" partitioned data. This article is an extended and consolidated version of a previous conference publication (Hartebrodt et al, 2021) with the following additional contributions: a demonstration how iterative leakage can pose a problem for federated power iteration; a further reduction in transmission cost, and increase in privacy, due to the use of randomized PCA; a data dependent speedup due to the use of approximate PCA. The remainder of this paper is organized as follows: In Section 2, we introduce concepts and notations that are used throughout the paper.…”

Section: Introductionmentioning

confidence: 99%

Federated singular value decomposition for high dimensional data

Hartebrodt¹,

Röttger²,

Blumenthal³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Federated learning (FL) is emerging as a privacy-aware alternative to classical cloud-based machine learning. In FL, the sensitive data remains in data silos and only aggregated parameters are exchanged. Hospitals and research institutions which are not willing to share their data can join a federated study without breaching confidentiality. In addition to the extreme sensitivity of biomedical data, the high dimensionality poses a challenge in the context of federated genome-wide association studies (GWAS). In this article, we present a federated singular value decomposition (SVD) algorithm, suitable for the privacy-related and computational requirements of GWAS. Notably, the algorithm has a transmission cost independent of the number of samples and is only weakly dependent on the number of features, because the singular vectors associated with the samples are never exchanged and the vectors associated with the features only for a fixed number of iterations. Although motivated by GWAS, the algorithm is generically applicable for both horizontally and vertically partitioned data.

show abstract