Topics

Variational approximation error in non-negative matrix factorization.

08:00 EDT 13th March 2020 | BioPortfolio

Summary of "Variational approximation error in non-negative matrix factorization."

Non-negative matrix factorization (NMF) is a knowledge discovery method that is used in many fields. Variational inference and Gibbs sampling methods for it are also well-known. However, the variational approximation error has not been clarified yet, because NMF is not statistically regular and the prior distribution used in variational Bayesian NMF (VBNMF) has zero or divergence points. In this paper, using algebraic geometrical methods, we theoretically analyze the difference in negative log evidence (a.k.a. free energy) between VBNMF and Bayesian NMF, i.e., the Kullback-Leibler divergence between the variational posterior and the true posterior. We derive an upper bound for the learning coefficient (a.k.a. the real log canonical threshold) in Bayesian NMF. By using the upper bound, we find a lower bound for the approximation error, asymptotically. The result quantitatively shows how well the VBNMF algorithm can approximate Bayesian NMF; the lower bound depends on the hyperparameters and the true non-negative rank. A numerical experiment demonstrates the theoretical result.

Affiliation

Journal Details

This article was published in the following journal.

Name: Neural networks : the official journal of the International Neural Network Society
ISSN: 1879-2782
Pages: 65-75

Links

DeepDyve research library

PubMed Articles [9419 Associated PubMed Articles listed on BioPortfolio]

Deep Non-Negative Matrix Factorization Architecture based on Underlying Basis Images Learning.

The non-negative matrix factorization (NMF) algorithm represents the original image as a linear combination of a set of basis images. This image representation method is in line with the idea of "part...

Kernelized Sparse Bayesian Matrix Factorization.

Extracting low-rank and/or sparse structures using matrix factorization techniques has been extensively studied in the machine learning community. Kernelized matrix factorization (KMF) is a powerful t...

Bayesian deep matrix factorization network for multiple images denoising.

This paper aims at proposing a robust and fast low rank matrix factorization model for multiple images denoising. To this end, a novel model, Bayesian deep matrix factorization network (BDMF), is pres...

Investigating microstructural variation in the human hippocampus using non-negative matrix factorization.

In this work we use non-negative matrix factorization to identify patterns of microstructural variance in the human hippocampus. We utilize high-resolution structural and diffusion magnetic resonance ...

Bright solitary waves as charge transport in DNA: A variational approximation.

Modeling energy and charge transfer in DNA has been a challenging issue because of many conformations DNA can take. Due to its simplicity, we propose a discrete variational approach to study the charg...

Clinical Trials [3359 Associated Clinical Trials listed on BioPortfolio]

Rectus Muscle Re-approximation at Cesarean Delivery

We include in our study 200 pregnant women, primigravidas, at term who underwent elective lower segment Cesarean section for the first time among the age group of 20 ̶ 35 years. The pati...

Direction Modulation of Muscle Synergies After a Stroke

The aim of the study is to investigate the capacity of post-stroke individual (study group) to modulate their EMG muscle activation pattern (MAP) compared to healthy individuals (control g...

The Effect of Rectus Muscle Approximation at Cesarean Delivery on Pain Perceived After Operation

This is a prospective randomized controlled study to determine the clinical outcome of rectus muscle re-approximation at Cesarean delivery.

The Papillary Muscle Approximation Provide Stability of Mitral Valve Repair for Ischemic Mitral Regurgitation

Recent publications show that an adjunctive subvalvular repair during mitral annuloplasty for secondary mitral regurgitation effective in preventing recurrent regurgitation. One of these p...

Cognitive Process of Diagnostic Error in Emergency Physicians

Diagnostic error, Dual process model of reasoning) During the last decade, much emphasis has been placed on system solutions to patient safety problems. However, diagnostic error, despite ...

Medical and Biotech [MESH] Definitions

A secreted matrix metalloproteinase that is the predominant proteolytic activity in the enamel matrix. The enzyme has a high specificity for dental enamel matrix protein AMELOGENIN.

A transmembrane domain-containing matrix metalloproteinase. It is synthesized as an inactive zymogen that is activated by the proteolytic action of PROPROTEIN CONVERTASES. Matrix metalloproteinase 16 plays a direct role in the cleavage of proteins in the pericellular environment. In addition it can function indirectly by enzymatically activating the proprotein form of other MATRIX METALLOPROTEINASES such as the zymogen of MATRIX METALLOPROTEINASE 2.

A secreted matrix metalloproteinase that is believed to play a role in EXTRACELLULAR MATRIX remodeling and cell fate determination during normal and pathological processes. Matrix metalloproteinase 11 was originally isolated in primary BREAST NEOPLASMS and may be involved in the process of tumorigenesis.

A transmembrane domain-containing matrix metalloproteinase. It is synthesized as an inactive zymogen that is activated by the action of PROPROTEIN CONVERTASES such as FURIN. Matrix metalloproteinase 14 plays a direct role in the cleavage of proteins in the pericellular environment. In addition it can function indirectly by enzymatically activating the proprotein form of MATRIX METALLOPROTEINASE 15.

The introduction of error due to systematic differences in the characteristics between those selected and those not selected for a given study. In sampling bias, error is the result of failure to ensure that all members of the reference population have a known chance of selection in the sample.

Quick Search


DeepDyve research library

Searches Linking to this Article