
Beta Process Nonnegative Matrix Factorization with Stochastic Structured MeanField Variational Inference
Beta process is the standard nonparametric Bayesian prior for latent fac...
read it

Prediction of Cancer Microarray and DNA Methylation Data using Nonnegative Matrix Factorization
Over the past few years, there has been a considerable spread of microar...
read it

AdversariallyTrained Nonnegative Matrix Factorization
We consider an adversariallytrained version of the nonnegative matrix f...
read it

Data embedding and prediction by sparse tropical matrix factorization
Matrix factorization methods are linear models, with limited capability ...
read it

On Contamination of Symbolic Datasets
Data taking values on discrete sample spaces are the embodiment of moder...
read it

GeNet: Deep Representations for Metagenomics
We introduce GeNet, a method for shotgun metagenomic classification from...
read it

Bayesian Tensor Filtering: Smooth, LocallyAdaptive Factorization of Functional Matrices
We consider the problem of functional matrix factorization, finding low...
read it
Doubly NonCentral Beta Matrix Factorization for DNA Methylation Data
We present a new nonnegative matrix factorization model for (0,1) boundedsupport data based on the doubly noncentral beta (DNCB) distribution, a generalization of the beta distribution. The expressiveness of the DNCB distribution is particularly useful for modeling DNA methylation datasets, which are typically highly dispersed and multimodal; however, the model structure is sufficiently general that it can be adapted to many other domains where latent representations of (0,1) boundedsupport data are of interest. Although the DNCB distribution lacks a closedform conjugate prior, several augmentations let us derive an efficient posterior inference algorithm composed entirely of analytic updates. Our model improves outofsample predictive performance on both real and synthetic DNA methylation datasets over stateoftheart methods in bioinformatics. In addition, our model yields meaningful latent representations that accord with existing biological knowledge.
READ FULL TEXT
Comments
There are no comments yet.