Warm-Starting in Message Passing algorithms

Skuratovs, Nikolajs; Davies, Mike E.

doi:10.1109/isit50566.2022.9834336

Cited by 2 publications

(1 citation statement)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, inspired by [31], we initialize the CG iteration with an estimate of the signal from the previous iteration of gVAMP. This warm-starting technique leads to a reduced number of CG steps that need to be performed and, therefore, to a computational speed-up.…”

Section: Methodsmentioning

confidence: 99%

Joint modelling of whole genome sequence data for human height via approximate message passing

Depope,

Bajzik,

Mondelli

et al. 2023

Preprint

View full text Add to dashboard Cite

Efficient utilization of large-scale biobank data is crucial for inferring the genetic basis of disease and predicting health outcomes from the DNA. Yet we lack accurate methods that scale to data where electronic health records are linked to whole genome sequence information. To address this issue, we develop a new algorithmic paradigm based on Approximate Message Passing (AMP), which is specifically tailored for genomic prediction and association testing. Our method yields comparable out-of-sample prediction accuracy to the state of the art on UK Biobank traits, whilst dramatically improving computational complexity, with a 10x-speed up in the run time. In addition, AMP theory provides a joint association testing framework, which outperforms the commonly used REGENIE method, in a third of the compute time. This first, truly large-scale application of the AMP framework lays the foundations for a far wider range of statistical analyses for hundreds of millions of variables measured on millions of people.

show abstract

Section: Methodsmentioning

confidence: 99%

Joint modelling of whole genome sequence data for human height via approximate message passing

Depope,

Bajzik,

Mondelli

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

Orthogonal Approximate Message-Passing for Spatially Coupled Linear Models

Takeuchi

2024

IEEE Trans. Inform. Theory

View full text Add to dashboard Cite

Orthogonal approximate message-passing (OAMP) is proposed for signal recovery from right-orthogonally invariant linear measurements with spatial coupling. Conventional state evolution is generalized to a unified framework of state evolution for the spatial coupling and long-memory case. The unified framework is used to formulate the so-called Onsager correction in OAMP for spatially coupled systems. The state evolution recursion of Bayes-optimal OAMP is proved to converge for spatially coupled systems via Bayes-optimal long-memory OAMP and its state evolution. This paper proves the information-theoretic optimality of Bayes-optimal OAMP for noiseless spatially coupled systems with right-orthogonally invariant sensing matrices.

show abstract

Warm-Starting in Message Passing algorithms

Cited by 2 publications

References 23 publications

Joint modelling of whole genome sequence data for human height via approximate message passing

Joint modelling of whole genome sequence data for human height via approximate message passing

Orthogonal Approximate Message-Passing for Spatially Coupled Linear Models

Contact Info

Product

Resources

About